CANCER-TESTIS GENE EXPRESSION AS A BIOMARKER OF THE

GENETIC VARIATION IN THE ONE CARBON METABOLIC PATHWAY

A THESIS SUBMITTED TO

THE DEPARTMENT OF MOLECULAR BIOLOGY AND GENETICS

AND THE INSTITUTE OF ENGINEERING AND SCIENCES OF

BILKENT UNIVERSITY

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR

THE DEGREE OF MASTER OF SCIENCE

BY

AHMET RASİM BARUTÇU

AUGUST 2008

II

I certify that I have read this thesis and that in my opinion it is fully adequate, in scope and in quality, as a thesis for the degree of Master of Science.

Assist. Prof. Dr. Ali Osmay Güre

I certify that I have read this thesis and that in my opinion it is fully adequate, in scope and in quality, as a thesis for the degree of Master of Science.

Prof. Dr. Bensu KARAHALİL

I certify that I have read this thesis and that in my opinion it is fully adequate, in scope and in quality, as a thesis for the degree of Master of Science.

Assist. Prof. Dr. Zeynep KALAYLIOĞLU-WHEELER

Approved for the Institute of Engineering and Science

Director of Institute of Engineering and Science

Prof. Dr. Mehmet Baray

III

CANCER-TESTIS GENE EXPRESSION AS A BIOMARKER OF THE

GENETIC VARIATION IN THE ONE CARBON METABOLIC PATHWAY

Ahmet Rasim Barutçu

MSc. in Molecular Biology and Genetics

Supervisor: Assist. Prof. Dr. Ali Osmay Güre

August 2008, 87 Pages

ABSTRACT

S-adenosyl methionine (SAM) is the sole methyl donor for all biological reactions in

humans. Folate consumption, required for SAM generation, is also essential for

dTMP synthesis and both events occur via enzymes of the one-carbon pathway.

Frequently occurring alleles of these enzymes have occasionally been associated

with several diseases including cancer. However, the cumulative effects of the

polymorphic variants of these enzymes on S-adenosylmethionine production have

not been studied. The identification of a biomarker that can reflect the collective

effect of these allelic variants is critical in moving the field forwards. We

hypothesized that Cancer-Testis (CT) genes, whose expression strongly correlates

with DNA hypomethylation, could be such a biomarker. In this study, we have

pursued an extensive correlation of CT expression and allelic variants of the several

one-carbon pathway enzyme genes , including methyltetrahydrofolatereductase

(MTHFR), methionine synthase (MS), reduced folate carrier (RFC) and methionine

synthase reductase (MTRR) in non-small cell lung cancer. Our results revealed

linkage disequilibrium among alleles as well as correlations between given

genootypes and CT gene expression, and illuminate the critical next steps that need

to be pursued.

IV

BİR KARBON METABOLİK YOLUNDAKİ GENETİK FARKLILIKLARIN BELİRLEYİCİSİ OLARAK KANSER-TESTİS GEN İFADESİ

Ahmet Rasim Barutçu

Moleküler Biyoloji ve Genetik Yüksek Lisansı

Tez Yöneticisi: Yard. Prof. Dr. Ali Osmay Güre

Ağustos 2008, 87 Sayfa

ÖZET

İnsanlarda bütün biyolojik tepkimelerde kullanılan tek metil grubu vericisi S-

adenozilmetiyonin (SAM)’dir. SAM üretimi için gereken folik asit tüketimi,

deoksitimidin monofosfat (dTMP) sentezi için gereklidir ve her iki olay da bir

karbon metabolik yolundaki enzimler sayesinde gerçekleşir. Bu enzimlerde sıkça

görülen aleller kanser dâhil birçok hastalık ile ilişkilendirilmiştir. Ancak, bu

enzimlerde görülen polimorfizmlerin SAM üretimine olan kümülatif etkisi şu ana

kadar çalışılmamıştır. Bir karbon metabolik yolundaki enzim alellerinin kümülatif

etkisine duyarlı ve buna cevap verebilecek biyolojik bir belirleyicinin teşhisi, bu

alanda ilerleme kaydedilmesi için önemlidir. Hipotezimize göre transkripsiyonları

DNA demtilasyonu ile ilişkili olan Kanser-Testis (KT) genleri, böyle bir biyolojik

belirteç olabilir. Bu çalışmada, küçük hücre dışı akciğer kanserinde (KHDAK) KT

gen ifadesi ile bir karbon metabolik yolundaki, metilentetrahidrofolat redüktaz

(MTHFR), metiyonin sentaz (MS), indirgenmiş folat taşıyıcısı (RFC) ve metiyonin

sentaz redüktaz (MTRR) enzim alelleri ilişkiyi araştırdık. Sonuçlarımız, enzim

alelleri arasında linkage disequilibrium olduğunu göstermiş, ayrıca belli bir haplotip

ile CT gen ifadesi arasındaki ilişkiyi açığa çıkarmış ve izlenecek adımlarda önemli

noktaları aydınlatmıştır.

V

TO MY FAMILY

VI

ACKNOWLEDGEMENTS

First of all, I would like to thank Assist. Prof. Dr. Ali Osmay Güre for his endless

support, supervision and guidance. I am grateful for his patience and this thesis could not

have been prepared without him. I feel very lucky and honored to work with Dr. Ali Güre

for I believe I have gained the maximum experience in my master study by working with

him.

We would never be able to finalize my results without the help of Assist. Prof. Dr.

Zeynep Kalaylıoğlu-Wheeler. I would like to thank her for the time and effort she has

dedicated. She performed critical statistical analyses without which interpretation of some of

our results might have been impossible. She has been a genuine support for my study.

I would also like to thank Prof. Dr. Bensu Karahalil for her valuable knowledge and

guidance. She graciously made precious contributions to my study.

I was delighted to have such a wonderful and amiable team of colleagues. They were

not only colleagues, but marvelous friends in and out of the lab. I will never forget our

monthly and weekly “meetings” and “feasts”. I would like thank Aydan Bulut, Duygu

Akbaş, Şükrü Atakan, Derya Dönertaş, Esen Oktay and Ender Avcı for their support. I do

not know if I will have such a laboratory team in my academic life. I would also like to

thank the MBG family for creating such a laboratory environment.

Last but never the least; I would like to thank my father for raising such an

individual and I wish he could see me making this work come true. I would like to thank my

dear family and my companion for their never-ending love and support.

During this the course of this study, towards my Master’s degree, I received a

Scholarship 2210 from TÜBİTAK.

7

Table of contents

ABSTRACT………………………………………………………………………………….....III

ÖZET……….…………………………………………………………………………….……..IV

DEDICATION PAGE…………………………………………………………………..……….V

ABSTRACT………………………………………………………………………………….....III

ÖZET……….…………………………………………………………………………….……..IV

DEDICATION PAGE…………………………………………………………………..……….V

ACKNOWLEDGEMENTS.................................................................................................... VI

TABLE OF CONTENTS ......................................................................................................... 7

LIST OF TABLES ................................................................................................................... 9

ABBREVIATIONS ............................................................................................................... 12

CHAPTER1. INTRODUCTION ............................................................................................ 13

1.1 DNA methylation ............................................................................................................. 13

1.2 DNA Methylation and Cancer .......................................................................................... 14

1.3 Cancer Testis (CT) Genes ................................................................................................. 15

1.3.1 CT Gene Expression Regulation ............................................................................ 16

1.4 S-Adenosylmethionine: The universal methyl donor......................................................... 19

1.5 One carbon pathway ......................................................................................................... 20

1.6 Folate deficiency and DNA damage ................................................................................. 23

1.7 One carbon pathway enzyme variants ............................................................................... 24

1.7.1 Methylenetetrahydrofolate reductase (MTHFR) ..................................................... 24

1.7.2 Methionine synthase reductase (MTRR) ................................................................ 26

1.7.3 Methionine synthase (MTR) .................................................................................. 26

1.7.4 Reduced Folate Carrier (RFC) ............................................................................... 27

1.7.5 Thymidylate synthase (TYMS) .............................................................................. 27

8

1.8 Cancer and one carbon pathway enzyme variants ............................................................. 28

1.9 The aim of the study ......................................................................................................... 31

CHAPTER 2. MATERIALS AND METHODS ..................................................................... 33

2.1 The PCR Method.............................................................................................................. 33

2.2 DNA Samples .................................................................................................................. 34

2.3 Restriction Fragment Length Polymorphism (RFLP) Analysis.......................................... 36

2.3.1 Digestion Products ................................................................................................. 37

2.4 c-DNA Synthesis .............................................................................................................. 43

2.5 Cell Culture ...................................................................................................................... 43

2.6 RNA Isolation .................................................................................................................. 43

2.7 Real Time PCR ................................................................................................................ 44

2.8 Statistical Analysis ........................................................................................................... 45

CHAPTER 3. RESULTS ....................................................................................................... 46

3.1 One carbon pathway enzyme genotype frequencies of lung cancer patients ...................... 46

3.2 Distribution of one carbon enzyme genotypes .................................................................. 48

3.3 One carbon enzyme allele distribution . ............................................................................ 50

3.4 CT expression associations with one carbon enzyme genotype combinations ................... 51

3.5 One carbon enzyme genotype associations in lung cancer patients .................................... 53

3.6 Univariate power analysis................................................................................................. 54

3.7 Multivariate power analysis .............................................................................................. 55

CHAPTER 4. DISCUSSION ................................................................................................. 58

AND FUTURE PERSPECTIVES .......................................................................................... 58

CHAPTER 5. REFERENCES ................................................................................................ 73

SUPPLEMENTARY FIGURES………………………………………..……………………...81

9

List of tables

Table 1.The one-carbon pathway enzyme allele activity differences. ...................... 28

Table 2. PCR primers used for RFLP analysis ........................................................ 33

Table 3. Nested PCR primers used for RFLP analysis ............................................ 34

Table 4. CT gene expression profile of CT (+) tumor samples ................................ 35

Table 5. CT gene expression profile of CT(-) tumor samples .................................. 36

Table 6. Expected RFLP product sizes ................................................................... 37

Table 8. A222V (C677T) variant of the MTHFR gene............................................ 38

Table 9. E429A (A1298C) variant of the MTHFR gene .......................................... 39

Table 10. D919G (A2756G) variant of the MTR gene ............................................ 40

Table 11. I22M (A66G) variant of the MTRR gene ................................................ 41

Table 12. R27H (G80A) variant of the RFC gene ................................................... 42

Table 13 One carbon enzyme distributions in lung cancer patients and Hardy-Weinberg

expectations. ........................................................................................ 47

Table 14. Distribution of 1-carbon enzyme genotypes among CT-positive and -negative lung

cancer patients I: Chi-square test. ......................................................... 49

Table 15.Distribution of 1-carbon enzyme genotypes among CT (+) and CT (-) lung cancer

patients II: Odds ratios ......................................................................... 50

Table 16. One carbon enzyme allele distribution among CT-positive and -negative lung cancer

patients. ............................................................................................... 51

10

Table 17. CT expression associations with 1-carbon enzyme genotype combinations in lung

cancer patients I: MTHFR677 C>T and RFC80 G>A. .......................... 52

Table 18. CT expression associations with 1-carbon enzyme genotype combinations in lung

cancer patients II: MTHFR1298 A>C and RFC80 G>A ....................... 53

Table 19. One carbon enzyme genotype associations in lung cancer patients ......... 54

Table 20. Univariate power analysis* ..................................................................... 55

Table 21. Multivariate power analysis* ................................................................. 56

Table 22. Genotype typing inconsistencies ............................................................. 59

Table 23. The microarray meta-analysis showing the average fold changes of gene expression for

the listed genes in cancer. ..................................................................... 70

Table 24 (Supplementary Table 1). The genotype data of the CT (+) lung cancer patients 81

Table 25 (Supplementary Table 2). The genotype data of the CT (-) lung cancer patients 82

11

List of Figures

Figure 1. CT gene expression is regulated by DNA methylation ............................. 18

Figure 2.The chemical structure of S-adenosylmethionine ...................................... 19

Figure 3. The one carbon pathway. ......................................................................... 21

Figure 4. RFPL analysis of RFC G80A polymorphism. .......................................... 60

Figure 5. An exemplary Q-PCR melting-curve analysis. ......................................... 62

Figure 6. A possible copy number variation in the MTRR A66G polymorphism .... 63

Figure 7. Proposed model I .................................................................................... 67

Figure 8. Proposed model II ................................................................................... 69

Figure 9. Q-PCR results showing the expression levels of TS ................................. 71

Supplementary Figure 1. RFLP results of the MTHFR C677T polymorphism ........ 83

Supplementary Figure 2. RFLP results of the MTHFR A1298C polymorphism. ..... 84

Supplementary Figure 3. RFLP results of the MTR A2756G polymorphism .......... 85

Supplementary Figure 4. RFLP results of the MTRR A66G polymorphism ............ 86

Supplementary Figure 5. RFLP results of the RFC G80A polymorphism…….....87

12

Abbreviations

DHF - Dihydrofolate

MTA- Methylthioadenosine

MTHFR- Methylenetetrahydrofolate reductase

MTR – Methionine synthase

MTRR - Methionine synthase reductase

RFC – Reduced Folate Carrier

RFLP – Restriction Fragment Length Polymorphism

SAM – S-adenosyl methionine

SAH – S-adenosyl homocysteine

SNP – Single nucleotide polymorphism

THF - Tetrahydrofolate

TS – Thymidylate synthase

13

CHAPTER1. INTRODUCTION

1.1 DNA methylation

The inheritance of information which is not based on DNA sequence is known as

epigenetics [1]. DNA methylation is a crucial determinant in gene expression, DNA stability and

chromatin modifications. The vitality of DNA methylation for vertebrates has not only been

shown in embryonic lethality of DNA methyl transferase-1 (DNMT1) knockout mice, but also

by its deregulation in various diseases, especially cancer [2]. DNA methylation is thought to be

evolved to silence viral sequences and transposable elements as well as to regulate gene

transcription [3]. DNA methylation is a heritable, tissue and cell specific modification of

cytosine residues in CpG sequences. The methyl group can be attached at the N4 or C5 positions

of the cytosine residues of prokaryotic or eukaryotic genomic DNA [4]. The distribution of the

CpG residues in the genome is not equal, some regions such as the Alu repeats and transcription

start sites contain a higher frequency of CpG residues than other regions [84].

Most of the mammalian genome consists of extended regions that are deficient for

CpG’s. CpG residues are at 20% of their predicted frequency in the mammalian genome. This is

thought to be a direct consequence of an increased mutation rate of the 5-methylcytosine residues

found in CpG sequences from cytosine to thymine. Most methylated CpG’s are found in clusters,

called a CpG island which corresponds to 1% of the human genome. The term CpG island refers

to a 500 base pair window with an increased G:C content of at least 55%, and an observed to

expected CpG frequency of at least 0.65. These islands may span the 5’ regions of

approximately half of the human genes including exons, promoters and untranslated regions.

Up to 80% of all cytosine residues, which are not related to CpG islands, are normally

methylated. In contrast, the CpG residues in CpG islands, especially at the gene promoter regions

of actively transcribed genes, are usually unmethylated. The major genomic regions which have

methylated CpG islands are the inactive X chromosome in female and silenced alleles of

parentally imprinted genes [5]. The CpG methylation also occurs at the sites with a low

frequency of CpG’s such as repeat DNA sites, heterochromatin, telomeres, non-coding regions

and exons. Methylation of the “bulk” of the genome enables the silencing of these non-coding

14

regions, which prevents the transcription of repeat elements and parasitic DNA sequences. DNA

methylation has also attracted a considerable interest in the cell differentiation and tissue specific

gene expression process [6].

DNA methylation is a dynamic process in which involves DNA methyltransferases

(Dnmt1, Dnmt3a and Dnmt3b), methyl-binding proteins, histone modifying enzymes, chromatin

remodeling factors and their molecular complexes [84]. It has been previously shown that

cyclical and dynamic methylation/demetylation of the CpG residues in a model containing the

ps2 gene promoter occurs with the presence of MeCP2 (a methyl-binding protein), SWI/SNF (a

histone remodeling complex), DNMT1 and DNMT3a and DNMT3b [7]. DNMT1 is responsible

for the maintenance methylation of the genome after each round of replication. During the

replication of eukaryotic genomic DNA, approximately 40 million CpG dinucleotides are

converted into the hemimethylated state in the newly synthesized DNA strand. These

hemimethylated CpG sites must be methylated precisely to maintain the original DNA

methylation pattern. DNMT1 is located at the replication fork and methylates newly

biosynthesized DNA strands directly after the replication. DNMT3a and DNMT3b

methyltransferases methylate CpG dinucleotides without preference for hemimethylated DNA,

and are responsible for the de novo methylation of DNA [8].

1.2 DNA Methylation and Cancer

For over a decade, abnormalities of DNA methylation in cancer cells have been

recognized [9]. In cancer, the usual pattern of methylation is observed to be genomic

hypomethylation and gene-specific hypermethylation. Hypermethylation, which is also related

to transcriptional silencing, is a common mechanism for the inactivation of several tumor

suppressor genes in cancer [10]. Methylations of such genes seem to occur early in

carcinogenesis, and in some cases increasing progressively leading to malignant phenotypes

[11]. Hypermethylation occurs not only at the promoter of tumor suppressor genes, but also at

the promoter regions of other genes involved in cancer progression such as DNA repair, cell

cycle regulation, apoptosis, hormonal response and cell adherence genes [12]. Previously, it has

15

been suggested that a tumor-type specific profile of DNA hypermethylation exists and thus

allows us to use these hypermethylated loci as biomarkers of tumorigenesis [13], [11].

On the other hand, global DNA hypomethylation is also seen in early carcinogenesis [14].

It has been previously shown that DNA hypomethylation causes genomic instability, thus

resulting in increased levels of DNA damage, mutation rates, copy number alterationd and loss of

heterozygosity. Chen et al have shown that hypomethylation in murine embryonic stem cells

lacking Dnmt1, at the same time significantly elevating mutation rates in two model genes [15]. It

has been suggested that there is a direct correlation between the methylation capacity of the cell

and genetic instability [16]. Abnormal DNA methylation has been associated with tumor

aggressiveness and poor prognosis [17]. Precancerous cells showing aberrant DNA methylation

profiles designate increased malignancy [18]. The reason for the instability of genomic DNA

observed along DNAhypomethylation might be the lack of a splice variant of Dnmt3b which

results in chromosomal instability through hypomethylation of pericentromeric satellite regions

[19]. In addition, Dnmt1 over-expression observed in cancer cells has been observed to associate

with increased CpG island methylation in a cancer-specific manner [19].

In normal cells, DNA methylation allows the condensation of the chromatin structure

through the recruitment of the chromatin-organizing proteins such as histone remodeling

complexes or polycomb group proteins [20]. In a DNA-hypomethylated cell, this arrangement is

lost which results in chromatin decondensation and chromosomal rearrangements [21].

1.3 Cancer Testis (CT) Genes

CT genes are a group of genes which are consistently expressed in spermatogonia,

oogonia and trophoblast cells, but not expressed in any other healthy tissues except cancer cells

[22]. CT genes can be categorized as families based on sequence similary. Some families such as

MAGE-A and SSX contain more than 8 members, while others like NY-ESO-1 consist of only

two. As will be explained below, despite the dissimilarity of their sequences, there is

overwhelming evidence that tumor-specific aberrant CT gene expression occurs in a coordinate

fashion. More than half of 40 CT genes, so far identified, are located on the X chromosome

[22],[23]. CT gene promoters (NY-ESO-1, MAGE-A1, SSX4 and SSX 7) lack a TATA box

16

[24], {unpublished data}. An interesting aspect of CT genes is that most of them are thought to

have risen due to duplications [23] and thus, they are located juxtameric to each other forming

gene clusters.

1.3.1 CT Gene Expression Regulation

Since CT genes are diverse, sequence-wise, it can be envisioned that they are possibly

regulated via unique transcription factors. Yet, the fact that all CT genes, studied to date, are co-

expressed, suggests the presence of a general mechanism behind their expression in cancer. DNA

methylation is thought to be the major control mechanism for the expression and silencing of CT

genes. This assumption was previously verified by a detailed analysis of the CpG islands of a

model CT gene, MAGEA1 [24]. It has been shown that the critical CpG islands at the promoter

region remain significantly methylated in vivo where the gene is not expressed [24]. Similary,

expression of all other CT genes, studied to date, correlates with the demethylation of their

promoter-proximal DNA regions [24](our unpublished data).

An example to such an experiment is detailed below SSX expression correlates with its

methylation status. The southern blot figure below shows the change in expression of SSX genes,

along with their methylation status (Figure 1). The general demethylation observed for the SSX

gene region seems to occur in parallel to L1 repeat demetylation, which in turn, is considered to

reflect the general methylation state of the genome.

The probes used in the assay were SSX 1, 2, 3, 4 and 5 (a mixture of SSX genes) and L1.

SSX probe recognizes all of the SSX exons whereas L1 probe recognizes all of the L1 repeat

sites. In this experiment, the genomic DNA from different cancer cell lines was cleaved by MspI

and HspII enzymes. Then, the digested products were separated by agarose gel electrophoresis.

Following the transfer of the DNA to the nitrocellulose membrane, the membrane wass exposed

to a hybridization probe, a single DNA fragment with a specific sequence whose presence in the

target DNA is to be determined. Then by autoradiography, the pattern of hybridization was

visualized.

17

Both MspI and HspII cut at CCGG, yet MspI enzyme cuts DNA regardless of the

methylation status while HspII cuts only the unmethylated DNA. Thus, the methylated SSX or

L1 sequence will not be digested by HspII, as seen for the Calu-3 cell line. On the other hand,

the unmethylated SSX or L1 sequences will be digested by HspII and, as observed for the SK-

LC17 cell line. . In fact, for this cell line, the digestion pattern observed for the two enzymes is

identical suggesting complete demetylation of SSX gene proximal CCGG sites. Below the

southern blot figure, semi-quantitative PCR results show that the expression of SSX genes

increase in parallel to the decrease of methylation within the SSX , as well as in L1 genomic

DNA.

18

Figure 1. Southern blot and PCR showing the expression levels along with their methylation status of SSX genes and L1 repeats. Probes: SSX1 and L1. M: Msp1, H: HspII

Thus, there are many examples demonstrating a clear relationship between

hypomethylation of genomic DNA and CT gene induction in cancer. CT genes can be

upregulated by DNA methyl transferase inhibitors [25] and histone deacetylase inhibitors as well

[22]. In 2005, Gure et al reported that in NSCLC (non-small cell lung cancer), CT gene

expression is frequently observed. In the same study, CT genes were found to be coordinately

expressed. Among 9 CT genes studied, expression of every single CT was found to correlate

significantly with any other [26].

Moreover, CT genes were found to be significantly associated with less differentiated,

higher grade of tumors, later stages of cancer and worse outcome. Larger tumor size and

invasion capacity are also correlated with CT expression [26]. It was also observed that CT

19

expression increases as the tumors progress [26]. Therefore, one can presume that the

hypomethylation observed in cancer cells causes demethylation in the promoter regions of CT

genes, thus increasing their expression. . Most importantly, in the same study, overall survival of

patients without (or with low levels of) CT gene expression was found to be significantly better

than those with high level CT gene expression in their tumors. Thus, given this relationship, if

the CT expression levels of tumors can be decreased, it can be hypothesized that this would

result in improved survival. In this study, I have addressed this question with reference to the

one-carbon pathway.

1.4 S-Adenosylmethionine: The universal methyl donor

S-adenosylmethionine (SAM) is a coenzyme involved in methyl group transfers. It is

synthesized from ATP and methionine by methionine adenosyltransferase. In this reaction, the

tri-phosphate group is cleaved from ATP and methionine is covalently attached. Figure 2 shows

the chemical structure of SAM.

Figure 2.The chemical structure of S-adenosylmethionine. The methyl group attached to the sulfur atom is reactive and is donated to an acceptor, forming S-adenosylhomocysteine., adapted from [27].

The methyl group attached to the sulfur atom is chemically reactive and thus allows the

donation of this methyl group to an acceptor substrate in the methylation reactions. The product

20

following these transmethylation reactions, is S-adenosylhomocysteine (SAH) is formed by the

demethylation of SAM.

Since SAM is the only methyl donor in the cell, the efficiency of SAM production might

possibly effect methylation reactions including DNA and histone methylation. A recent study

showed that rats fed with a methyl-deficient diet displayed significantly reduced levels of SAM,

reduced ratios of SAM to SAH (SAM’s more stable metabolite) and that this correlated with

elevated levels of DNA hypomethylation [28]. Since CT gene expression is controlled by DNA

methylation, a deficiency in SAM production could in turn be crucial in the regulation of CT

genes. The low production of SAM might result in DNA hypomethylation, causing CT up-

regulation and poor prognosis in cancer. For this reason, we have focused on the one carbon

pathway in order to investigate the possible mechanisms which consequently results in reduced

SAM production and DNA hypomethylation.

1.5 One carbon pathway

One carbon metabolism is an intervention of two pathways enabling the cross-talk

between epigenetic and genetic processes, which involves DNA methylation and DNA synthesis.

One carbon metabolism is vitally important for the maintenance of methionine cycle, nucleotide

synthesis and biological methylation reactions. Folate is the most important input substrate

utilized in the one carbon pathway. Folate is not only an essential co-factor for the de novo

biosynthesis of purines and pyrimidines, it also plays an important role in DNA synthesis,

stability and integrity [29]. Figure 3 summarizes those particular events in the one carbon

pathway which we chose to focus on.

21

Diet

Serum Folate RFC

Folic Acid

DHF

THF

5-10-methylene THF

5-methylene THF

MTHFR

Methionine

Hcy

MTRMTRR B12

SAM

Biological

Methylation

reactions

SAH

dUMP

dTMP

TS

DNA synthesis

Figure 3. The one carbon pathway. Annotations: MTHFR- Methylenetetrahydrofolate reductase, MTRR-

Methionine synthase reductase, MTR- Methionine synthase, RFC- Reduced Folate Carrier, TS- Thymidylate

synthase, DHF- Dihydrofolate, THF-Tetrahydrofolate, SAM- S-adenosylmethionine, SAH- S-

adenosylhomocysteine, Hcy- homocysteine

Folate may be gathered from two sources; either from foods or from blood by serum folate.

Because folate cannot pass through the cell membrane when its glutamate tail is longer than 3, it is

absorbed in the small intestine after the hydrolysis of polyglutamate chain by glutamate

carboxypeptidase II (GCPII) [30]. Two receptors can transport folate from blood to into the cells.

The folate receptor (FR), and the reduced folate carrier (RFC). The FR has a higher affinity for

oxidized folate when compared to RFC. When taken into the cell, folate is first converted to

dihydrofolate, and then to tetrahydrofolate (THF) by the enzyme dihydrofolate reductase (DHFR).

The following reaction generates the origin of one carbon units by the breakage of β-carbon of

22

serine. In this reaction catalayzed by serine hydroxymethyltransferase (SHMT), THF is converted to

N5-N10-methylene-THF while glycine is produced. N5-N10-methylene-THF is a crucial intermediate

factor for the direction of the pathway [30].

For thymidine synthesis, deoxythymidylate monophosphate (dTMP) is synthesized from

deoxyuridylate monophophate (dUMP) by thymidylate synthase (TS). TS transfers a methyl group

from N5-N10-methylene-tetrahydrofolate. The non-reversible methylation of dUMP to dTMP results

in the oxidation of N5-N10-methylene-THF to the inactive dihydrofolate, which can be converted

back to THF by DHFR [31].

For de novo methionine biosynthesis and methylation reactions, methylene

tetrahydrofolate reductase (MTHFR) converts N5-N10-methylene-THF to 5-methyl-THF. This

step creates the only source of 5-methyl-THF in the one carbon pathway. Methionine synthase

(MTR or MS), a cobalamin dependent enzyme which is activated by methionine synthase

reductase (MTRR or MSR), transfers one methyl group from 5-methyl-THF in order to convert

homocysteine (Hcy) into methionine. Dimethyl glycine is also produced from choline and

betaine for methionine generation. Methionine synthesis ensures the provision of the universal

methyl donor S-adenosyl methionine (SAM), with the activation of methionine by methionine

adenosyl transferase and ATP. SAM is used to methylate more than 80 important biomolecules

such as DNA, RNA, and proteins including histones. During this process, SAM is converted into

S-adenosyl-homocysteine (SAH), which is either hydrolyzed to homocysteine (Hcy) to initiate a

new remethylation cycle or transsulphurated to cysteine by cystathionine β-synthase [30].

In a cell, every methylation process requires a methyl group transfer from SAM. All of

the metabolic methylation reactions are under the effect of SAM, which directs the utilization of

Hcy and indicates the level of methylation. As explained before, the SAM/SAH ratio in a cell

determines the cellular methylation potential. SAH is known to be an inhibitor of

methyltransferases [23]. The level of SAM in the cell is regulated by two mechanisms. First of

all, SAM suppresses the synthesis of N5-methylene-THF by inhibiting the MTHFR enzyme.

Thus, methionine level and therefore Hcy level decreases when SAM increases. On the other

hand, homocysteine can also be converted to methionine through SAH hydrolase. Therefore, N5-

23

methylene-THF is an important molecule determining the pathway between remethylation and

transsulphuration reactions. If there is sufficient N5-methylene-THF entering the methylation

pathway, the direction of the SAH favors cysteine synthesis; if not, SAH is used to produce

methionine [32]. The source of N5-methylene-THF is folate acquired by diet. Folate, as well as

B12 and B6 vitamins, which function as co-enzymes for MTR and cysthation synthase,

respectively, regulate the homocysteine removal and in turn, prevent SAH accumulation [33].

1.6 Folate deficiency and DNA damage

There is accumulating evidence that folate deficiency, either due to low dietary folate

intake or due to low blood folate levels, is related to tumorigenesis [1]. Perturbations of the one

carbon metabolism are thus, thought to play vital roles in neoplastic development by affecting

gene regulation through DNA methylation and genome integrity through DNA synthesis and

repair [34].Folate deficiency has been shown to induce hepatocellular carcinoma development in

rats [1]. Diets deficient in various methyl donor groups (folic acid, choline, methionine and

vitamin B12) have been shown to induce DNA hypomethylation, site-specific DNA

hypermethylation, double strand breaks, and upregulation of DNMT’s [1].

Low dietary folate intake is strongly associated with DNA damage through uracil

misincorporation or by DNA hypomethylation leading to genome instability [35]. Decreased

cytosolic levels of N5-N10-methylene-THF caused by low folate status decreases the dTMP

synthesis and increase the dUMP/dTMP ratio [35]. The increase of uracil pool in the cell elevates

the rate of DNA-polymerase-mediated dUTP misincorporation into the DNA [36]. Uracil is

excised from DNA by uracil-DNA glycosylase and apyrimidinic endonuclease, generating nicks

which will be ligated by DNA ligase. However, if these nicks (single strand breaks) are located

on two opposite strands, less repairable and hazardous double strand breaks may form. Uracil

mediated double strand breakage is the major cause of deletions, duplications, chromosome

breaks, micronucleus formation, chromatid recombinations, fragile sites and translocations which

play important roles in tumorigenesis [35]. In cell culture, folate depletion can increase the

24

intracellular dUMP/dTMP ratio up to 10 fold. Folate depletion induced DNA breakage has been

observed in lymphocytes and rodent cell lines in vitro [37], [38].

Decreasing folate levels can cause upregulation of DNA damage and cell cycle

checkpoint related genes such as p53, p16 and p21 as well [37]. The normal levels of dNTPs for

normal DNA/RNA synthesis are directly dependent on intracellular folate availability.

Maintenance of folate concentration from diet or supplements is correlated with a protective

effect and reduced incidence of number of cancers [39].

On the other hand, genomic instability caused by global hypomethylation is characterized

by an elevated rate of DNA sequence changes, aneuploidy, chromosome translocations and gene

amplifications [40]. A balance between the two critical reactions (methylation and DNA

synthesis/repair) must be reached so one does not compromise the other [41].

1.7 One carbon pathway enzyme variants

Table 1 summarizes studies where polymorphic variants of the one carbon pathway

enzyme activities were investigated utilizing various assays. From the enzymatic activity data

obtained from the literature, one can infer that the variations of the enzymes partaking in SAM

generation can create disturbances in the methylation process.

1.7.1 Methylenetetrahydrofolate reductase (MTHFR)

All four enzymes which I focused on in this study have allelic variants in the population

and are associated with cancer. Different studies have focused on each of these alleles

individually, despite the possibility that the net cumulative effect of various alleles is what

ultimately determines SAM production efficiency. A large number of allelic variants of the one

carbon pathway enzymes described. For example, there are 65 SNPs found on the MTHFR.

However, among these, only 10 are gene non-synonymous SNPs that are found in more than

25

10% of the population are only 2 for MTHFR [42]. For some of these alleles, an association with

decreased enzymatic activity has been described [42], (Table 1). Cancer risk generally has been

linked to such hypomorphic alleles of one carbon pathway enzymes [33]. The major variant of

the MTHFR gene, C677T transition (rs1801133), resulting in an Ala to Val change in aa.222,

was found to decrease the activity of the MTHFR enzyme by %70 [43], (Table 1). This common

variant of the MTHFR gene is associated with an increased risk of cancer [44]. MTHFR 677 TT

homozygosity causes a significant increase in the homocysteine levels and DNA

hypomethylation when compared to the wild-type allele [45].

Apart from the C677T transition, A1298C transition (rs1801131), converting Glu 429 to

Ala, of the MTHFR enzyme causes a 15% reduction in the activity and results in increased levels

of plasma homocysteine levels [45], Table 1. The impaired activity of MTHFR, especially along

with the low folate status, results in less production of N5-methylene-THF and methionine, thus

proceeds to hypomethylation [45]. In another study, carriers of A1298C polymorphism do not

appear to have high levels of plasma Hcy (Table 1) but have a lesser degree of MTHFR activity

reduction compared to C677T variants, when tested by biochemical assays [46]. However, when

combined with other variants, A1298C polymorphism becomes effective in determining the Hcy

and folate levels. Individuals heterozygote for both alleles for both the C677T and A1298C

variants have significantly increased plasma Hcy levels when compared to 677 CC / 1298 AA

homozygotes [47]. Although in several studies, the 677TT variant of MTHFR was found

preferentially associated with cancer [48], in other studies, this variant was found to reduce

cancer risk by directing the pathway to thymidine synthesis and thus preventing uracil

accumulation [49]. In the MTHFR gene, as high as 65 SNP’s exist in a population of 240

individuals. Interestingly, the combinations of different polymorphisms in the gene coding

MTHFR result in an additive effect of reduction in the enzyme activity [42]. So, it becomes an

important aspect to study the polymorphisms collectively to overcome the possibility of missing

the effect of a polymorphism that can be observed only when within the context of a particular

genotype.

26

1.7.2 Methionine synthase reductase (MTRR)

Methionine synthase reductase (MTRR), an intermediate methyl carrier during the

remethylation of Hcy to methionine, may be another important determinant of SAM production.

In the past decade, an A to G polymorphism (A66G) in the 66th base pair (rs1801394), resulting

in an isoleucine to methionine substitution in the 22nd aminoacid, was found to be associated

with neural tube defects and cancer, especially in the absence of B12 vitamin and in the presence

of the MTHFR C677T variant [48],[50]. In one study where a biochemical assay was used,

demonstrated that the 66 GG variant of the MTRR enzyme had only 25% enzymatic activity

compared to the wild-type protein [51], (Table 1). This low enzymatic activity results in an

inefficient production of SAM [52]. This is likely due to the decreased capacity of MTRR to

activate MTR, in turn would lead to impaired production of methionine from Hcy. Not

independently, but when combined with the 677T MTHFR variant, a cell carrying the MTRR

66G variant will have a lower capacity of remethylation, which will direct to DNA

hypomethylation, instability and damage [48].

1.7.3 Methionine synthase (MTR)

A variation in the methionine synthase (MTR) enzyme (A2756G), which results in a

subsitution of aspartic acid to glycine in the 919th aminoacid, also decreases the rate of

methylation of Hcy to methionine (rs1805087). Although an increased plasma Hcy concentration

is not observed in neural tube defect patients carrying the hypomorphic allele, it is thought to

have an additive effect along with other variants of other enzymes, and especially the MTRR

variants [53]. This aspartic acid to glycine substitution occurs in the vicinity of the binding

domain of vitamin B12 of the MTR [54]. It was observed that plasma homocysteine levels are

lower in MTR 2756 GG individuals when compared to the more common AA genotype [48],

(Table 1).

27

1.7.4 Reduced Folate Carrier (RFC)

Another important variant in the one carbon pathway is in the reduced folate carrier

(RFC) enzyme. The RFC gene codes an integral membrane protein which intakes the folate into

the cell. A guanine to adenine transition (rs1051266) at the 80th base pair (G80A) of RFC is

related to a moderate degree of elevated plasma Hcy and plasma folate levels [55]. This arginine

to histidine transition results in a higher affinity of RFC for folate impairing its subsequent

intracellular release [55]. A recent study pointed out that this polymorphism reduces N5-methyl-

tetrahydrofolate transport efficiency to 54% compared to the wild-type allele [56], (Table 1).

1.7.5 Thymidylate synthase (TYMS)

The enhancer region of the thymidylate synthase gene contains a number of 28-base-pair

tandem repeats. These repeats are either two repeats or three repeats, while three repeats

occurring most frequently. The triple repeat has been associated with a 2.6 fold increase in the

expression of TYMS [57]. One study showed that 3rpt/3rpt subjects have lower plasma folate

and higher plasma Hcy levels than those with other genotypes [58]. This tandem repeat

polymorphism, combined with reduced folate intake, has been associated with lung

adenocarcinomas, [59]. Conversely, among those individuals with the 3rpt/3rpt genotype, higher

folate intake was correlated with a 50% reduced risk of cancer [60].On the other hand, in

individuals carrying a 2rpt/2rpt genotype, higher folate intake is correlated with a 50% increased

cancer risk [60].Similar results were shown for vitamin B12, but not with vitamin B6, methionine

or alcohol intake [60].

28

Table 1.The one-carbon pathway enzyme allele activity differences*.

Enzymatic activity Assay Genotypes Reference

CC CT TT

MTHFR C677T Micobiological 100% 65% 30% [53]

Biochemical 100% 55% [42]

RBC folate 100% 90% 82% [61]

RBC folate 100% 90% 112% [45]

Plasma folate 100% 96% 75% [45]

Total Hcy 100% 101% 121% [55]

Total Hcy 100% 108% 123% [45]

DNA hypomethylation 100% 98% 121% [45]

CpG Hypermethylation 100% 82% 28% [62]

MTHFR A1298C

AA AC CC

Biochemical 100% 102% [42]

Plasma folate 100% 112% 114% [46]

Plasma folate 100% 110% 107% [45]

Total Hcy 100% 75% 77% [46]

Total Hcy 100% 90% 115% [45]

Genomic DNA methylation 100% 231% 182% [46]

DNA hypomethylation 100% 102% 119% [45]

CpG Hypermethylation 100% 107% 24% [62]

MTR A2756G

AA AG GG

CpG Hypermethylation 100% 28% 3% [62]

AA AG GG

MTRR A66G Biochemical 100% 25% [51]

GG GA AA

RFC G80A Plasma (total) Hcy 100% 103% 124% [55]

5-CHO-mH4F transport 100% 57% [56]

*Bold numbers show differences that were found to be statistically significant

1.8 Cancer and one carbon pathway enzyme variants

Epidemiologic studies where these alleles were studied individually and in pairs,

suggested that they could be associated with cancer. Additive and synergistic effects of one

carbon pathway enzyme variants and related dietary factors, such as folic acid and methionine

29

consumption, are complex and in order to clarify the association of each allelic variant with

cancer, these, as well as their interactions with dietary factors need to be studies carefully. On the

basis of functional effects of these polymorphic variants, there ought to be a correlation between

the incidence of cancer and the hypoactive alleles of these enzymes.

However, there seems to be an inverse association between the MTHFR 677TT and

cancer, especially in situations with sufficient folate intake [49]. A study has shown that the

2rpt/2rpt phenotype of TYMS and MTHFR 677 TT together result in a statistically decreased

risk of hepatocellular carcinoma [41], MTHFR 677TT and 1298CC, together, are also associated

with a reduced risk of cancer [41]. It seems likely that in folate deficiency, MTHFR 677TT

becomes almost totally incapable of producing N5-methylene-THF and thus, these cells have an

impaired methylation capacity [52]. On the other hand, even when there is sufficient folate in the

one carbon pool, the hypomorphic MTHFR protein will not be able to convert N5-N10-

methylene-THF effectively; so, TYMS will use it for thymidine synthesis. One can envision that

because this will prevent the accumulation of uracil and therefore, prevent uracil

misincorporation, the “T” allele will protect against tumorigenesis by avoiding DNA damage

caused by dUMP. In contrast, the TT allele was also shown to increase the cancer risk by 2.64

fold in folate deficiency and 1.6 fold in folate sufficiency [44]. Friso et al have shown that the

MTHFR 677 TT allele associates with a significantly lower level of methyl cytosine in DNA,

when compared to 677 CC and CT tissues, but only under conditions of low folate status [63]. At

higher folate levels, the methyl cytosine levels did not differ from that among the 677 “CC”

individuals [63]. That the 677 TT individuals have low levels of methyl cytosine in their DNA

and elevated levels of plasma Hcy suggests that the MTHFR TT allele results in insufficient

amounts of N5-methylene-THF that can’t meet the demands for the de novo methylation. In

another study, the MTHFR 677TT and the MTRR 66GG variant combination showed the highest

amount of DNA damage, calculated by micronucleus formation [47]. Vaughn et al showed that

the MTRR alleles have no effect on plasma folate and plasma vitamin B12, in the presence of a

677 CC or CT allele. However, in those individuals homozygous for the 677 T allele, the

existence of MTRR AG or GG allele resulted in a significant increase in plasma homocysteine

30

levels [64]. This observation suggests that the functional MTRR 66 A/A allele compensates for a

hypoactive MTHFR 677 TT allele.

It is worthwhile to point out that the association of MTHFR 677 variants is different in

different stages and types of cancer. While the 677 CC and CT alleles are related to a reduced

risk in renal cell carcionoma [65], they are related to an elevated risk in colorectal cancer [48]].

The effect may be different at different stages of the oncogenesis process. Supporting this is the

observation that folate deficiency predisposes to colon cancer, but once neoplastic lesions are

present, folate supplementation actually accelerates colon cancer transformation {Kim, 2003}.

Folate supplementation after a certain stage of tumorigenesis may enhance DNA synthesis in an

already transformed cell.

In a study implemented by Chango et al, the RFC allele does not seem to affect plasma

homocysteine levels when the MTHFR 677 allele is CC or CT. However, in the presence of the

677 TT allele, RFC GA and AA will influence the utilization of folate [55].

It is a challenging task to determine whether uracil misincorporation or errors in

methylation process causes oncogenesis. Moreover, how the genes in the one carbon metabolism

respond to folate deficiency is largely unknown. For instance, in the HCT116 cell line, folate

deficiency appeared to preferentially shuttle the flow of one carbon units to the methionine cycle

to protect methylation reactions and thereby suppress DNA synthesis. However in Caco2 cells in

the same experimental setup, the metabolic priority in response to folate depletion was to shuttle

the available folate pools to the nucleotide biosynthesis pathway at the expense of the methionine

cycle [31].

The studies of the one carbon pathway enzyme variants suggest that folate has an

important role in modulating epigenetic features of the DNA that controls gene expression [66].

Since the end product of the one carbon pathway, S-adenosylmethionine, will be utilized for the

de novo methylation of DNA, the efficiency of the production of SAM will affect gene

expression via DNA methylation. Folate depletion is thought to cause tumorigenesis either by

impairing methylation or by hindering DNA synthesis {Kim, 2003}. When the effects of one the

31

carbon metabolism enzymes are combined with the effects of low folate intake, significant

results on the alterations of DNA methylation are obtained suggesting the roles of both events.

.

1.9 The aim of the study

The role of DNA methylation in differentiation, disease and tissue specific gene

expression has received considerable attention. In many circumstances, DNA methylation can on

its own be an effective mechanism for gene silencing [24]. Methylated silent genes can be re-

activated by using demethylating agents [67]. It has been proposed that the methylation of CpG

sites at gene promoter regions inhibits the recruitment of the transcription machinery and thus

strongly represses the expression [24]. DNA methylation is proposed to be the basic mechanism

for the regulation of CT genes such as MAGE [24] and SSX (Gure, 2005}. Interestingly,

spermatogonia cells undergo epigenetic reprogramming, which involves genomic

hypomethylation [68].

Since genomic hypomethylation causes an increase in CT expression which is associated

with developed stages of cancer and poor outcome (Gure, 2005}, the reversal of the genomic

hypomethylation might improve prognosis. Because the methylation process requires the

effective production of SAM by the one carbon pathway, we decided to investigate the

relationship between CT gene expression and the allelic variants of the one carbon pathway,

which we think are the primary effectors of SAM production.

It is already known that a variety of enzyme variants in the one carbon pathway cause

perturbations in the methylation cycle. Several epidemiologic studies have shown that variants of

one carbon pathway enzymes are correlated with oncogenesis in different ways [69]. One can

envision that the perturbations of the one carbon pathway hindering the methylation capacity of

the cell may result in an additive effect resulting in DNA hypomethylation and might lead thus,

to CT positivity and poor prognosis. We hypothesized that there could be a correlation between

the allelic variants of the one carbon pathway enzymes and CT expression. The importance of

such correlation would be that if CT positivity can be predicted by the allele variants a cancer

32

patient, the possible conversion of high CT expression to a low state by the addition of dietary

supplements such as folic acid could improve the prognosis of the patient.

One can envision that typing for the one carbon enzyme alleles among individuals who

have a high-risk of developing cancer might help us suggest precautions towards restoring the

patient’s impaired methylation capacity, such as the fortification or the supplementation of the

diet by folic acid.

In this study, I determined the one carbon pathway enzyme genotypes of lung cancer

tumors and cell lines, and studied the correlation of these genotypes with CT expression levels.

The allelic variants I chose to study were the non-synonymous, highly prevalent and for which

there were a large number of epidemiologic studies indicating a correlation with cancer. The

selected variants include MTHFR C677T, MTHFR A1298C, MTRR A66G, MTR A2756G and

RFC G80A single nucleotide polymorphisms.

I, thus, determined the one carbon enzyme genotypes of individuals who were previously

typed for CT expression, using the restriction fragment length polymorphism assay (RFLP). In

this method, the region containing the single nucleotide polymorphism (SNP) is amplified by

PCR. The PCR product is subsequently digested with a specific endonuclease, which can

recognize and cut DNA wherever a specific short sequence exists. This process is called a

restriction digestion. This method can only be used if the allelic variant can be distinguished by a

specific restriction endonuclease. Thus, restriction enzymes were selected such that the PCR

product of one allele would be digested whereas the other allele would not. When the digestion

products are analyzed by agarose gel electrophoresis, the allele is determined according to the

molecular weights of the digestion products.

In summary, apart from many epigenetic mechanisms which may be perturbed in cancer,

DNA hypomethylation seem to be the major actor for the ectopic expression of CT genes. By

elucidating the underlying mechanisms which might thus cause CT expression, we potentially

will have described a novel biomarker useful for the determination of tumorigenesis as well as

improving the prognosis and lifespan of a cancer patient.

33

CHAPTER 2. MATERIALS AND METHODS

2.1 The PCR Method

The standart reaction mixture contained 1µg of genomic DNA, 1µl of 10mM dNTP mix

(Finnzymes, Cat. No: F-506L), 2µl of 25mM primer mix, 2.5 µl of 10X Buffer (Finnzymes

Dnazyme II HotStart Reaction Buffer, Cat. No: F-522), 0.5 µl Hot Start Taq Polymerase

(Finnzymes, Cat. No: F-504L) added up to 25 µl with ddH2O. Then, the tubes are transferred to

the Perkin Elmer (PE9700) thermal cycler for 35 cycles. Before the reaction, initial denaturation

was performed at 94 oC for 10mins. Each cycle consisted of 30sec at 94 oC, 30sec at Tm oC, and

30 sec at 72 oC. Then, the mixture was incubated at 72 oC for 7mins for final annealing and a

final hold of 4 oC. See Table 2 for primer pairs, Tm values and Table 3 for nested primers.

Table 2. PCR primers used for RFLP analysis

Gene Name and SNP Gene ID Ref. Seq. Primers Tm Product Size(bp)

MTHFR C677T 4524 AY338232.1 5’ – TTTGAGGCTGACCTGAAGCAC – 3’ (sense) IIF 3’ – GACCTGAGAGGAGATCTGG – 5' (antisense) IIR

60C 288 (8,700nt-8,988nt)

MTHFR A1298C 4524 AY338232.1 5’ – GAGGAGCTGCTGAAGATGTG -3’ (sense) IIF 3’- TGGAGGTCTCCCAACTTACC -5’(antisense) IIR

65C 261 (10,605nt-10,866nt)

MTR A2756G 4548 NT_004836 5’- TGTTATCAGCATTGACCATTACTACAC -3’(sense) IIF 3’- ACTGTTTCAGCACCTGTTTCCC -5’ (antisense) IIR

65C 498 (105,261nt-105,759nt)

MTRR G66A 4552 NT_006576 5’ - GCAAAGGCCATCGCAGAAGACAT -3’(sense) IF 3’- CACTTGTTCTCACAGCCACCC -5’(antisense) IIR

60C 380 (7,860,967nt-7,861,347nt)

RFC G80A 6573 AL163302.2 5’ – CTCCCGCGTGAAGTTCTT -3’ (sense) IF 3’ – AGCGTCACCTTCGTCCCCTC -5’ (antisense) IIR

60C 231 ( 236,569nt-236,800nt)

TS Polymorphism 7298 NC_000018.8 5' - GTTCCCGGGTTTCCTAAGAC -3' (sense) IIF 3'- GATCTGCCCCAGGTACTGCA -5' (antisense) IIR

65C 390

TS Expression 7298 NC_000018.8 5' - GCAGATCCAACACATCCTCC - 3' (sense) 3' - CCATTGGCATCCCAGATTTTCAC -5' (antisense)

60C 236

34

Table 3. Nested PCR primers used for RFLP analysis Gene Name and

SNP Ref. Seq. Nested Primers Product size (bp)

nested Tm

MTHFR C677T AY338232.1 5’- TGAAGGAGAAGGTGTCTGCGGGA – 3’ (sense nested) IF 3’- AGGACGGTGCGGTGAGAGTG -5’ (antisense nested) IR

198 (8723nt-8921nt) 60oC

MTHFR A1298C AY338232.1 5’ – GAGGAGCTGACCAGTGAAG -3’(sense nested) IF 3’- GGTAAGTTGGGAGACCTCCA -5’ (antisense nested) IR

136 (10.629nt-10.765nt) 60oC

MTR A2756G NT_004836 5’- TGTTCCCAGCTGTTAGATGAAAATC -3’(sense nested) IF 3’- GATCCAAAGCCTTTTACACTCCTC – 5’(antisense nested) IR

211(105.334nt-105.545nt)

65oC

MTRR G66A NT_006576 3’-GTGAAGATCTGCAGAAAATCCATGTA -5’(antisense nested) IR* 66bp (7,860,967nt-7,861,032nt)

60oC

RFC G80A AL163302.2 5’- AGCCGTAGAAGCAAAGGTAGC – 3’(sense nested) IIF 3’ – TGCATTCGTCTCCAGGGTG – 5’(antisense nested) IR

154bp (236,646nt-236,667nt)

55oC

TS NC_000018.8 5'-AGCAGGAAGAGGCGGAGC-3'(sense nested) 3'- CCGGCCACAGGCATG -5'(antisense nested)

172bp 60oC

*The 5’(forward) nested primer used for MTRR was identical to that shown in Table 2.

2.2 DNA Samples

Tumor samples from 763 lung cancer patients were previously collected and typed for

CT expression (Gure, 2005). All of the samples were scored for the positivity of CT expression.

The scores corresponding to different expression levels of each CT gene ranged between a

negative (-) to three positives (+++). We then selected, genomic DNA prepared from these tumor

samples with the highest and the lowest amount of CT expression and used them for one carbon

pathway allele genotype analysis. Table 4 shows the label, cancer type and the CT expression

scores of the samples. Table 4a and Table 4b the CT(+) and CT (-) lung cancer samples used to

for this study respectively.

35

Table 4. CT gene expression profile of CT (+) tumor samples

Lu CT(+) HISTOLOGY NY-ESO-1 LAGE-1 MAGE-A1 MAGE-A3 MAGE-A4 MAGE-A10 CT-7 SSX-2 SSX-4

31 SQCC (+++) (+/-) (+) (+++) (+++) (+)

68 Adeno (+++) (+) (+) (-)

87 Adeno (-) (-) (+++) (+++) (+++)

89 Adeno (+++) (+++) (+++) (+++) (+/-) (+) (+/-)

111 Lgcell (-) (-) (+++) (+++) (+++) (+++)

131 SQCC (+++) (+/-) (+) (+++) (+++)

168 SQCC (+++) (+++) (+++) (+++) (+++) (+)

185 NSCLC (+++) (+++) (+++) (+++) (+/-) (-) (-) (++) (-)

186 Adeno (+++) (+++) (+++)

219 SQCC (+) (+) (+++) (+++) (+++) (-) (+/-) (+++)

223 SQCC (+++) (+++) (+++) (-)

649 AdenoBAC (+++) (+/-) (+++) (+++) (+/-) (+++) (-) (-)

652 SQCC (+++) (+++) (+++) (+++) (+++) (+++) (-) (-)

658 AdenoBAC (+/-) (+++) (+++) (+++) (+++) (+/-) (+++)

726 Adeno (+++) (+++) (+++) (+++) (+++) (+++) (++) (+++)

736 Adeno (+++) (+++) (+++) (+++) (+++) (+++) (+/-) (-)

739 Lgcell (+++) (+++) (+++) (+++) (+++) (+++) (++) (+++)

745 SQCC (+++) (+++) (+++) (+++) (+++) (+/-) (++) (-)

752 (+++) (+++) (+++) (+++) (+++) (+/-) (-) (-)

753 (++) (+/-) (+++) (+++) (+++) (+/-) (-)

759 (+++) (+++) (+++) (+++) (+++) (+/-) (-)

36

Table 5. CT gene expression profile of CT(-) tumor samples

Lu CT(-) HISTOLOGY NY-ESO-1 LAGE-1 MAGE-A1 MAGE-A3 MAGE-A4 MAGE-A10 CT-7 SSX-2 SSX-4

69 BAC (-) (-) (-) (-) (-)

77 SQCC (-) (-) (-) (-) (-)

88 Adeno (-) (-) (-) (-) (-)

90 Adeno (-) (-) (-) (-) (-)

108 Adeno (-) (-) (-) (-) (-)

110 Adeno (-) (-) (-) (-) (-)

112 SQCC (-) (-) (-) (-) (-)

180 Adeno (-) (-) (-) (-) (-) (-) (-)

183 BAC (-) (-) (-) (-) (-) (-) (-)

191 Adeno (-) (-) (-) (-) (-) (-)

221 Adeno (-) (-) (-) (-) (-)

225 Adeno (-) (-) (-) (-) (-) (-) (-) (-)

639 AdenoBAC (-) (-) (-) (-) (-) (-) (-)

656 AdenoBAC (-) (-) (-) (-) (-) (-)

657 SQCC (-) (-) (-) (-) (-) (-)

670 AdenoBAC (-) (-) (-) (-) (-) (-)

692 AdenoBAC (-) (-) (-) (-) (-) (-) (-) (-)

693 (-) (-) (-) (-) (-) (-) (-) (-)

694 (-) (-) (-) (-) (-) (-) (-) (-)

698 Adeno (-) (-) (-) (-) (-) (-) (-) (-)

706 Adeno (-) (-) (-) (-) (-) (-) (-) (-)

707 Adeno (-) (-) (-) (-) (-) (-) (-) (-)

713 AdenoBAC (-) (-) (-) (-) (-) (-) (-) (-)

716 Adeno (-) (-) (-) (-) (-) (-) (-) (-)

718 AdenoBAC (-) (-) (-) (-) (-) (-) (-) (-)

728 AdenoBAC (-) (-) (-) (-) (-) (-) (-) (-)

748 AdenoBAC (-) (-) (-) (-) (-) (-) (-) (-)

749 Adeno (-) (-) (-) (-) (-) (-) (-) (-)

751 AdenoBAC (-) (-) (-) (-) (-) (-) (-) (-)

763 (-) (-) (-) (-) (-) (-)

2.3 Restriction Fragment Length Polymorphism (RFLP) Analysis

37

RFLP analysis was done as previously described [43]. Following the PCR reaction, 5 µl

of PCR product was mixed with 1 µl of restriction endonuclease enzyme (New England

Biolabs), 1 µl of Buffer 2 (New England Biolabs), added up to 10 µl with ddH2O. The mixture

was incubated at 37oC overnight. Table 4 shows the expected sizes of the digestion products for

each primer pair.

Table 6. Expected RFLP product sizes

Polymorphism and Restriction endonuclease

Primers

IIF-IIR IIF-IR IF-IIR IF-IR

MTHFR C677T C/C (HinF1I) 288 220 266 198

MTHFR C677T C/T (HinF1I) 288, 241, 47 220, 173, 47 266, 241, 25 198, 173, 25

MTHFR C677T T/T (HinF1I) 241, 47 173, 47 241, 25 173, 25

MTHFR A1298C C/C (MboII) 266 160 237 136

MTHFR A1298C C/A (MboII) 261, 209, 28, 24 160, 108, 28, 24 237, 209, 28 136, 108, 28

MTHFR A1298C A/A(MboII) 209, 28, 24 208, 28, 24 209, 28 108, 28

MTR A2756G A/A (HaeIII) 498 284 425 211

MTR A2756G A/G (HaeIII) 498, 345, 153 284, 153, 131 425, 345, 80 211, 131, 80

MTR A2756G G/G (HaeIII) 345, 153 153, 131 345, 80 131, 80

MTRR G66A G/G (Nde1) - - 381 66

MTRR G66A G/A (Nde1) - - 381, 317, 39, 25 66, 42, 24

MTRR G66A A/A (Nde1) - - 317, 39, 25 42, 24

RFC G80A A/A (HinP1I) 287 154 364 231

RFC G80A A/G (HinP1I) 287, 257, 30 154, 124, 30 364, 257, 70, 37 231, 124, 70,37

RFC G80A G/G (HinP1I) 257, 30 124, 30 257, 70, 37 124, 70, 37

2.3.1 Digestion Products

38

2.3.1.1 MTHFR C677T

The MTHFR C677T polymorphism creates a HinF1I restriction site. Table 6 shows the

transition of the protein and the nucleotide sequence and the digestion products of all alleles.

Upon digestion, HinF1I will be able to cut the “T” allele whereas the “C” allele will remain

uncut.

Table 7. A222V (C677T) variant of the MTHFR gene

• Amino acid sequence A222V (Ref Seq: NM_005957):

151- KNIMALRGDP IGDQWEEEEG GFNYAVDLVK HIRSEFGDYF DICVAGYPKG

201- HPEAGSFEAD LKHLKEKVSA GADFIITQLF FEADTFFRFV KACTDMGITC

251- PIVPGIFPIQ GYHSLRQLVK LSKLEVPQEI KDVIEPIKDN DAAIRNYGIE

• Nucleotide Sequence C677T (Ref Seq: NM_005957):

GANTC – HinF1I restriction site

781 aggccacccc gaagcaggga gctttgaggc tgacctgaag cacttgaagg agaaggtgtc

841 tgcgggagcc gatttcatca tcacgcagct tttctttgag gctgacacat tcttccgctt

901 tgtgaaggca tgcaccgaca tgggcatcac ttgccccatc gtccccggga tctttcccat

961 ccagggctac cactcccttc ggcagcttgt gaagctgtcc aagctggagg tgccacagga

1021 gatcaaggac gtgattgagc caatcaaaga caacgatgct gccatccgca actatggca

C/T T/T C/C

39

2.3.1.2 MTHFR A1298C

Upon the transition of adenine to cytosine at the 1298th position of the MTHFR gene, the

recognition site of the MboII enzyme becomes abolished. Therefore, the undigested products will

indicate the presence of an “C” allele whereas the “A” allele will result in digestion. The amino

acid and the nucleotide alterations are designated in Table 7.

Table 8. E429A (A1298C) variant of the MTHFR gene

Amino acid sequence E429A (Ref Seq: NM_005957):

351- LSAHPKRREE DVRPIFWASR PKSYIYRTQE WDEFPNGRWG NSSSPAFGEL

401- KDYYLFYLKS KSPKEELLKM WGEELTSEES VFEVFVLYLS GEPNRNGHKV

451- TCLPWNDEPL AAETSLLKEE LLRVNRQGIL TINSQPNING KPSSDPIVGW

• Nucleotide Sequence A1298C (Ref Seq: NM_005957):

GAAGA…N8 –MboII restriction site

1321 ggagtgggac gagttcccta acggccgctg gggcaattcc tcttcccctg cctttgggga

1381 gctgaaggac tactacctct tctacctgaa gagcaagtcc cccaaggagg agctgctgaa

1441 gatgtggggg gaggagctga ccagtgaaga aagtgtcttt gaagtcttcg ttctttacct

1501 ctcgggagaa ccaaaccgga atggtcacaa agtgacttgc ctgccctgga acgatgagcc

1561 cctggcggct gagaccagcc tgctgaagga ggagctgctg cgggtgaacc gccagggcat

1621 cctcaccatc aactcacagc ccaacatcaa cgggaagccg tcctccgacc ccatcgtggg

C/C A/C A/A

40

2.3.1.3 MTR A2756G

The A to G transition of this polymorphism creates a HaeIII restriction site. Thus, the G

allele will be able to be digested. The information about the transition is indicated in Table 8.

Table 9. D919G (A2756G) variant of the MTR gene

• Amino acid sequence D919G (Ref Seq: NP_000245):

781 kgdvhdigkn ivgvvlgcnn frvidlgvmt pcdkilkaal dhkadiigls glitpsldem

841 ifvakemerl airiplligg attskthtav kiaprysapv ihvldasksv vvcsqllden

901 lkdeyfeeim eeyedirqdh yeslkerryl plsqarksgf qmdwlsephp vkptfigtqv

961 fedydlqklv dyidwkpffd vwqlrgkypn rgfpkifndk tvggearkvy ddahnmlntl

• Nucleotide Sequence A2756G (Ref Seq: NC_000001):

GGCC – HaeIII restriction site

89641 attgaccatt actacaccag ttttatcatc ttttgctcat ctatggctat cttgcatttt

89701 cagtgttccc agctgttaga tgaaaatcta aaggatgaat actttgagga aatcatggaa

89761 gaatatgaag atattagaca ggaccattat gagtctctca aggtaagtgg tagaaacaga

89821 tttttgcttg tttttaatgt gactgttttt tatgatccta gtttttaatg tgacttttta

89881 aaatggtttt gaggagtgta aaaggctttg gatcatttta gagaatttct gtcttctagt

41

2.3.1.4 MTRR A66G

The A66G transition in the MTRR gene, upon with a substitution of a base with modified

primers, creates a restriction site for the NdeI enzyme. The “A” allele will produce digested

products where as the “G” allele will remain uncut.

Table 10. I22M (A66G) variant of the MTRR gene

• Amino acid sequence I22M (Ref Seq: NP_002445):

1 mrrflllyat qqgqakaiae eiceqavvhg fsadlhcise sdkydlktet aplvvvvstt 61 gtgdppdtar kfvkeiqnqt lpvdffahlr ygllglgdse ytyfcnggki idkrlqelga

121 rhfydtghad dcvglelvve pwiaglwpal rkhfrssrgq eeisgalpva spassrtdlv

• Nucleotide Sequence A66G (Ref Seq: NC_000005):

CATATG – Nde1 restriction site gcaaaggccatcgcagaagcaat -primer with a modified base

1681 gccttgaagt gatgaggagg tttctgttac tatatgctac acagcaggga caggcaaagg

1741 ccatcgcaga agaaatatgt gagcaagctg tggtacatgg attttctgca gatcttcact

1801 gtattagtga atccgataag gttagagccg ttacagtgga ttttaccgtt ttgtgctttg

1861 aagaattttg gttgggaagt gatatttatg aaacaaaagg acactaatac caccacatag

1921 tctttgtttt ttaacagaaa tgtgtttgtt caatggtata gtaagatatc accagcattt

42

2.3.1.5 RFC G80A

The G to A transition at the 80th base of the RFC gene abolishes a HinP1I restriction site.

So, the G allele will be recognized and cut by HinP1I whereas the A allele will be undigested.

Table 11. R27H (G80A) variant of the RFC gene

• Amino acid sequence R27H (Ref Seq: NP_919231):

1 mvpsspavek qvpvepgpdp elrswrhlvc ylcfygfmaq irpgesfitp yllgpdknft

61 reqvtneitp vlsysylavl vpvflltdyl rytpvlllqg lsfvsvwlll llghsvahmq

121 lmelfysvtm aariayssyi fslvrparyq rvagysraav llgvftssvl gqllvtvgrv

• Nucleotide Sequence G80A (Ref Seq: NC_000021):

GCGC – HinP1I restriction site

4441 ccttcgtccc ctccggagct gcacgtggcc tgagcaggat ggtgccctcc agcccagcgg

4501 tggagaagca ggtgcccgtg gaacctgggc ctgaccccga gctccggtcc tggcggcacc

4561 tcgtgtgcta cctttgcttc tacggcttca tggcgcagat acggccaggg gagagcttca

4621 tcacccccta cctcctgggg cccgacaaga acttcacgcg ggagcaggca tgtgggtgcc

43

2.4 c-DNA Synthesis

c-DNA synthesis was carried out with a c-DNA synthesis kit (Finnzymes DyNAmo

cDNA synthesis kit, #F-470L). The master mix contained 10 µl of 2X RT Buffer, 1 µl of random

hexamer primer set with a concentration of 300µg/µl, 2 µ l of M-MuLV Rnase H+ reverse

transcriptase, maximum 1 µg of template RNA added up to 20 µl with ddH2O. After transferring

to the Perkin Elmer (PE9700) thermal cycler, the mixture was incubated in the following

conditions: 25 oC for 10 mins, 37 oC for 30mins, 85 oC for 5 mins and a final hold of 4 oC.

2.5 Cell Culture

Cells were grown at a confluence of approximately 80% in RPMI with 10% fetal calf

serum (Hycole), supplemented with glutamine, streptomycin and penicillin. The old medium was

removed and the cells were washed with 1X PBS. After removal of the 1X PBS, Trypsin/EDTA

solution was added and the cells were left for incubation at 37C for 3 - 4 minutes. When the cells

were detached from the surface of the plate, trypsin was inactivated by adding fresh growth

medium. The detached cells were suspended using a pipettor and were placed in a falcon tube.

The cells then were centrifuged at 800rpm for 5mins and the supernatant containing the

Trypsin/EDTA solution was removed. The pellet was re-suspended with fresh medium and cell

suspension was plated to a new flask. Fresh medium is added to the flask and placed in 37 oC,

%5 CO2 incubator.

2.6 RNA Isolation

Cells were trypsinized as previously described in the Cell Culture section. Cold media is

added and the cells are mixed. The cells are transferred into a 15ml falcon tube and centrifuged

at 800rpm for 5 mins, at 4 oC. After centrifugation, the media was removed and washed with 1X

PBS and 1ml Trizol (Invitrogen) was added. The cells were mixed until they homogenize and

transferred into an eppendorf tube. 200µl chloroform was added and the tube was vigorously

shaken for 15 seconds. After incubating for 3 minutes at room temperature, the cells were

44

centrifuged at 13.000rpm for 15 mins, at 4 oC. After this step, the supernatant was removed by a

pipettor and placed into a new tube. 500 µl isopropanol was added and the supernatant is mixed.

The mixture was incubated for 10 mins at room temperature and centrifuged at 13.000 rpm for

10 minutes, at 4 oC. Then the isopropanol was completely removed and 1ml 75% EtOH was

added following 8 minutes of centriguation at 8000rpm, at 4 oC. The 75% EtOH wassucked and

1ml 99.8% EtOH was added following the same centrifugation step of 8 mins at 8000rpm, 4 oC.

99.8% EtOH was removed and the pellet is dried under laminar flow for 5-10mins. Lastly, the

pellet is dissolved in 20µ l-25µl DNase-RNase free H2O and mixed

2.7 Real Time PCR

The real-time RT-PCR assays were done with the iCycler instrument (BioRad

Laboratories) using the Finnzymes Dynamo SYBR Green qPCR kit (Finnzymes Cat #F-410).

The primers used for TS expression were 5' -GCAGATCCAACACATCCTCC-3' (sense) and 3'-

CCATTGGCATCCCAGATTTTCAC-5' (antisense). β-actin was used as a loading control. The

PCR reactions were set up in a volume of 20 µl, containing 2 µl of sample cDNA, 10 µl of 2X

Master mix, 5 pmol from each TS specific primers, added up to 20 µl with RNase-DNase free

water. The cycling conditions were as follows: 95 °C for 1 min, 58 °C for 1 min, and 72 °C for

1min for 40 cycles with initial melting at 95 °C for 10 min.

Relative expression levels were calculated using the PCR threshold cycle number (CT)

for each sample and control sample (both of which were normalized according to β-actin mRNA

for differences in amount of total RNA added to the reaction), using the formula

2−(∆CT

sample−∆CT

control) [70], [71]. ∆CT represents the difference in CT values between the target and

β-actin transcripts. RT-PCR was performed in triplicates for each sample and average CT values

were calculated

45

2.8 Statistical Analysis

For the significance tests used in the percentage tables, chi-square test was used. The

significance level was determined as p=0.05. For the sample number calculation tests, univariate

logistic regression model analysis (Wald’s test) was performed by the Power and Precision

software. Multivariate logistic regression models were done by applying an algorithm in R

software [83]. For both of the tests the significance level was set to α=0.05. The event rate is

defined as the probability of a genotype being CT (+). The power of the analysis is defined as the

ability of the test to differentiate the difference between the alleles in terms of CT expression.

The odds ratio in multivariate tests indicates the ratio of the probability of a genotype being a CT

(+) sample divided by the probability of the same genotype being a CT (-) sample and the

division of the probabilities of the other genotypes being CT (+) and CT (-) samples,

respectively. An exemplary formula showing this ratio for the MTHFR 677 polymorphism is

indicated below. The estimate values in the multivariate test is calculated as log(OR).

46

CHAPTER 3. RESULTS

3.1 One carbon pathway enzyme genotype frequencies of lung cancer patients

All 50 lung cancer samples were typed for five allelic variants, except for MTR A2756G

and MTRR G66A, for which 1 and 4 samples could not be typed, respectively. We observed a

frequency of 44 % CC, 34% CT and 11% TT for the MTHFR C677T polymorphism, 48% AA,

46% AC and 6% CC for MTHFR A1298C, 52% AA and 48% AG for MTR A2756G, 38.1%

AA, 47.6% AG and 14.3% GG for MTRR A66G and 22% GG, 56% AG and 22% GG genotypes

for RFC G80A polymorphisms..

Next, we checked to see whether our lung cancer sample panel was in Hardy-Weinberg

equilibrium. For this reason, we calculated the expected frequencies of each genotype and

evaluated chi-square analysis for observed/expected ratio. The genotype frequencies for all the

polymorphisms were in accordance with the Hardy–Weinberg equilibrium: MTHFR C677T

(p=0.4), MTHFR A1298C (p=0.7), MTR A2756G (p=0.08), MTRR A66G (p=1) and RFC G80A

(p=0.9). Table 12 summarizes our observations and expected values for each genotype.

47

Table 12 One carbon enzyme distributions in lung cancer patients and Hardy-Weinberg expectations.

Polymorphism Genotype Observed Observed/

Expected

χ2 value p*

MTHFR677 CC 21 1.1 CT 19 0.8 TT 10 1.3 2.0 0.4

MTHFR1298 AA 24 1 AC 23 1.1 CC 3 0.7 0.7 0.7

MTR2756 AA 25 0.9 AG 24 1.3 GG 0 - 5.2 0.08 Undetermined (1)

MTRR66 GG 16 1 GA 20 1 AA 6 1 0.004 1 Undetermined (4)

RFC80 GG 11 0.9 GA 27 1.1 AA 12 0.9 0.3 0.9

*Chi square

48

3.2 Distribution of one carbon enzyme genotypes among CT (+) and CT (-) lung cancer

patients I

Among CT (+) samples the distribution was 44% CC, 26% CT and 30 % TT, while CT (-)

samples’ distribution was 31% CC, 41% CT and 28% TT.

The distribution of other polymorphisms was not significantly different when samples

were stratified for CT gene expression. The MTHFR A1298C polymorphism, the distribution

was 57% AA, 38% AC and 5% CC for CT (+) lung cancer samples and, 41% AA, 52% AC and

7% CC among the CT (-) samples.

The observed genotype frequencies for the MTR A2756G polymorphism were 63% AA

and 37% AG among the CT (+); 55% AA and 45% AG among the CT (-) samples. For the

MTRR A66G polymorphism, our observation for the CT (+) lung cancer samples was a

frequency of 35% AA, 55% AG and 10% GG genotypes. On the other hand, CT (-) lung cancer

samples were 35% AA, 46% AG and 19% GG.

We found 29% GG, 48% AG and 24% AA genotype frequency in CT (+) samples for the

RFC G80A polymorphism. Among the CT (-) samples, the distribution was 17% GG, 59% AG

and 24% AA.

49

Table 13. Distribution of 1-carbon enzyme genotypes among CT-positive and -negative lung cancer patients I:

Chi-square test.

Polymorphism Genotype CT(+) CT(-) p* MTHFR677 CC 12 (44%) 9 (31%) CT 7 (26%) 12 (41%) TT 2 (30%) 8 (28%) 0.02 MTHFR1298 AA 12 (57%) 12 (41%) AC 8 (38%) 15 (52%) CC 1 (5%) 2 (7%) 0.2 MTR2756 AA 10 (63%) 16 (55%) AG 11 (37%) 13 (45%) GG 0 (0%) 0 (0%) 0.3 MTRR66 GG 7 (35%) 9 (35%) GA 11 (55%) 12 (46%) AA 2 (10%) 5 (19%) 0.3 Undetermined 4 RFC80 GG 6 (29%) 5 (17%) GA 10 (48%) 17 (59%)

AA 5 (24%) 7 (24%) 0.2

*Chi square

Next, for each polymorphism, we grouped heterozygotes with the homozygote where

both showed a similar bias towards a given CT expression phenotype, and thus generated 2 by 2

charts. The odds ratio calculated from these charts demonstrated borderline significance for the

MTHFR677 polymorphism, with an odds ratio of 2.96, when other polymorphisms, when thus

evaluated, were not significantly different when CT(+) and (-) samples were compared (Table

14). Both analyses suggest that the hypoactive MTHFR677 CT/TT phenotypes preferentially

associate with lack of CT gene expression. .

50

Table 14.Distribution of 1-carbon enzyme genotypes among CT (+) and CT (-) lung cancer patients II: Odds

ratios

Polymorphism Genotype CT(+) CT(-) OR (95% CI) p* MTHFR677 CC 12 9 CT/TT 9 20 2.96 (0.92-9.53) 0.07 MTHFR1298 AA 12 12 AC/CC 9 17 1.89 (0.61-5.89) 0.3 MTR2756 AA 10 16

AG/GG 11 13 0.85 (0.28-2.61) 0.6 MTRR66 GG 7 9 GA/AA 13 17 0.9 (0.26-3.2) 0.8 Undetermined 4 RFC80 GG 6 5 GA/AA 15 24 1.92 (0.5-7.41) 0.3

*Fisher’s exact test (2-sided)

3.3 One carbon enzyme allele distribution among CT-positive and -negative lung cancer patient.

Since the previous analyses suggested an association between the MTHFR677 CT and

TT genotypes with the lack of CT gene expression, we tested whether a similar association could

be found for a given allele. As shown in Table 15, the hypoactive T allele of the MTHFR677

polymorphism was strongly associated with the lack of CT gene expression as well. We did not

observe a significant association between CT gene expression and the presence of any other

allele (Table 15).

51

Table 15. One carbon enzyme allele distribution among CT-positive and -negative lung cancer patients.

Polymorphism Allele CT(+) CT(-) OR (95% CI) p* MTHFR677 C 31 11 T 30 28 2.63 (1.11-6.21) 0.025 MTHFR1298 A 32 10 C 39 19 1.56 (0.64-3.82) 0.3 MTR2756 A 31 11 G 45 13 0.81 (0.32-2.05) 0.7

MTRR66 G 25 15 A 30 22 1.22 (0.53-2.84) 0.6 RFC80 G 22 20 A 27 31 1.26 (0.57-2.8) 0.6

*Fishers exact test (2-sided)

3.4 CT expression associations with one carbon enzyme genotype combinations in lung cancer

patients

Since combinations of genotypes are known to result in inefficient utilization of folate, as

well as decreased enzymatic activity, we tested whether a given combination of two genotypes

would result in a significant stratification of samples based on CT gene expression. Among all

possible two-by-two combinations, we found borderline significance between the lack of CT

52

expression and the presence of both the MTHFR677 “CC or TT” genotype, and the RFC80 “GG

or AA” genotype (Table 16); as well as when the MTHFR1298 “AC or CC” genotype was

combined with the RFC80 “GG or AA” genotype (Table 17). This is in line with what we

observed for individual genotype associations with CT gene expression, therefore, that

hypomorphic genotypes in the 1-carbon pathway associate with the lack of CT gene expression

in tumors.

Table 16. CT expression associations with 1-carbon enzyme genotype combinations in lung cancer patients I:

MTHFR677 C>T and RFC80 G>A.

RFC80

MTHFR677 CT(+) CT(-) OR (95% CI) p*

GG CC 8 9 1.9 (0.51-7.05) 0.05 GG CT/TT 7 15

GA/AA CC 4 0 - 0.3 GA/AA CT/TT 2 5

*Fisher’s exact test (2-sided)

53

Table 17. CT expression associations with 1-carbon enzyme genotype combinations in lung cancer patients II:

MTHFR1298 A>C and RFC80 G>A

RFC80

MTHFR1298 CT(+) CT(-) OR (95% CI) p*

GG AA 2 4 0.13 (0.01-2) 0.2 GG AC/CC 4 1

GA/AA AA 10 8 4 (1.01-15.71) 0.06 GA/AA AC/CC 5 16

*Fishers exact test (2-sided)

3.5 One carbon enzyme genotype associations in lung cancer patients

We tested whether the presence of genotype correlated with that of another. When all

lung cancer samples were evaluated, we observed a significant negative correlation between the

MTHFR677 CC and MTRR66 GG genotypes, between the MTHFR1298 AA and MTHFR66

GG genotypes, and co-occurrence of RFC80 “GG or GA” genotype with MTHFR1298 AA, as

well as with MTHFR677 CC (Table 18). A positive correlation was also present between the

RFC80 AA genotype and MTRR2756 AA. When samples were stratified according to CT gene

expression, the correlation between RFC80 AA and MTRR2756 AA genotypes was only

observed among CT(+) samples (Table 18). Among the CT(-) samples, the RFC80 “GG or GA”

correlation with MTHFR1298 AA was observed, in addition to a novel negative correlation

between MTRR66 GG and RFC80 GA (Table 18).

54

Table 18. One carbon enzyme genotype associations in lung cancer patients

Correlation Coefficient (Spearman’s rho) Genotype 1 Genotype 2 All CT(+) CT(-) p* MTHFR677 CC MTRR66 GG -0.417 N.C.& N.C. 0.006 MTHFR1298 AA MTRR66 GG -0.354 N.C N.C. 0.02 MTHFR1298 AA RFC80 GG/GA 0.45 N.C. 0.482 0.001/0.01# MTRR2756 AA RFC80 AA 0.338 0.63 N.C. 0.02/0.001

MTHFR677 CC RFC80 GG/GA 0.281 N.C. N.C. 0.05 MTRR66 GG RFC80 GA N.C. N.C. -0.452 0.03

*Fishers exact test (2-sided); &N.C.: no correlation; #p values correspond to the first and second correlation coefficients, respectively.

3.6 Univariate power analysis of CT expression and one carbon enzyme genotype associations

We then calculated the minimum sample size we would required to differenciate samples

according to their CT gene expression status, if genotypes were studied as two groups. Using the

Power and Precision software, we performed Wald’s test and a univariate logistic regression

model. It is important to point out that the polymorphisms are analyzed independently, regardless

of the effect of other variants studied. Our analysis revealed that as few as 80 samples would

help us distinguish samples by their CT expression pattern if they were classified into MTFF66

AA and “AG or GG” genotype groups. Similarly, including less than 200 samples in a

prospective study is predicted to help distinguish samples according to their RFC80 as well as

MTHFR677 polymorphisms (Table 19). It is important to point out that the polymorphisms are

analyzed independently, regardless of the effect of other variants studied.

55

Table 19. Univariate power analysis*

Genotype Event rates** Sample size MTRR66 AA vs. AG/GG 0.20 vs. 0.51 80 RFC80GG vs. GA/AA 0.67 vs. 0.42 128 MTHFR677 CC vs. CT/TT 0.29 vs. 0.51 159 MTHFR1298 AA vs. AC/CC 0.56 vs. 0.41 352

MTHFR2756 AA vs. AG/GG 0.45 vs. 0.50 3135

* Logistic regression model, Wald’s test, α=0.05, power=80; **Probability of a sample being CT

(+) for the given genotype

3.7 Multivariate power analysis of CT expression and one carbon enzyme genotype

associations

The univariate power analysis model assumes that the distribution of genotypes of

polymorphisms other than that for which the calculations are made occur at random frequencies.

However, we already determined that genotypes co-exist among lung cancer samples, as well as

within CT(+) and CT(-) groups, as explained above. We wanted to determine the minimal

sample number that could distinguish CT (+) from CT(-) samples for a given genotype would be,

if the others behaved similar to our observations. We, therefore, performed a multivariate power

analysis for each genotype controlling for the others. The minimum sample sizes calculated are

shown in Table 20. The beta variant in this table is defined as the logarithm of the odds ratio

defined on page 47. Thus, the MTRR66 AA versus AG/GG genotype distribution has a

coefficient of -1.43 and requires at least 132 samples to distinguish CT (+) and CT (-) samples,

given the genotypes of other enzymes are held constant as indicated (Table 20) . The beta

variant in this table is defined as the logarithm of the odds ratio defined on page 47. Thus, the

MTRR66 AA versus AG/GG genotype distribution has a coefficient of -1.43 and requires at

56

least 132 samples to distinguish CT (+) and CT (-) samples, given the genotypes of other

enzymes are held constant as indicated (Table 20) .

Table 20. Multivariate power analysis*

Genotype Adjusting for Beta Sample size

MTRR66 AA

MTRR66 AG/GG

MTHFR677=CC, MTHFR1298=AA,

MTR2756=AA, RFC80=GG

-1.43 132

MTHFR677 CC

MTHFR677 CT/TT

MTHFR1298=AA, MTR2756=AA, MTRR66=AA, RFC80=GG

1.02 162

RFC80GG

RFC80GA/AA

MTHFR677=CC, MTHFR1298=AA,

MTR2756=AA, MTRR66=AA

1.03 165

MTHFR1298 AA

MTHFR1298 AC/CC

MTHFR677=CC, MTR2756=AA,

MTRR66=AA, RFC80=GG

0.61 480

MTHFR2756 AA

MTHFR2756 AG/GG

MTHFR677=CC, MTHFR1298=AA,

MTRR66=AA, RFC80=GG

-0.2 5253

* Logistic regression model, Wald's test, α=0.05, power=80

Note that there is a difference between the sample sizes found in the univariate and the

multivariate tests. This finding constitutes the backbone of our study, which argues that it is

important to study these alleles comprehensively in order to overcome the possibility of missing

the effect of a polymorphism which is disregarded.

57

From the results derived from the multivariate analysis mentioned above, we have, with

an 80% power, calculated the standard errors and p-values in the condition where the sample size

is large enough to find a significant difference. For example, if we studied ≥132 samples, we

would be able to distinguish the CT (+) and CT (-) lung cancer samples by testing for MTRR 66

AA and AG/GG genotypes.

These statistical analyses give us an idea about the size of the sample and the genotypes

to be studied in our future experiments. For instance, since the MTR 2756 genotype requires

5253 samples, it is very likely that we would not include the typing of this genotype in a larger

epidemiologic study. Besides, we have observed that when the combined effects of the

polymorphisms are taken into consideration, we obtain a different result when they are not

collectively analyzed and this indicates the importance of studying these alleles collectively.

58

CHAPTER 4. DISCUSSION

AND FUTURE PERSPECTIVES

The study conducted by Gure et al in 2005 revealed that CT genes were coordinately

expressed and associated with poor prognosis [26]. Data from the same group and others showed

that CT gene expression was primarily regulated by DNA methylation [24]. The outcome of

these studies constituted the basis of the current study. S-adenosylmethionine is the one and only

methyl donor in the cell and therefore it is likely that an event altering SAM production would

affect CT expression. The polymorphisms in the one carbon pathway enzyme genes, which are

thought to cause reductions in SAM production, have already been shown to correlate with DNA

hypomethylation and increased SAH levels. Therefore, we decided to type lung cancer samples

whose CT gene expression status had been identified previously for the one carbon pathway

enzyme variants and check whether a given haplotype could predict the existence of CT

expression so that further studies for silencing the CT expression to improve the prognosis can

be conducted.

In this study, the typing of the polymorphisms was conducted by the restriction fragment

length polymorphism (RFLP) method. In order to verify our results, we tried to reproduce our

data by a number of methods. While a total of 127 experiments were performed only once, 36

experiments were repeated two times and 13 experiments were repeated three times or more. We

found inconsistent results for a total of 9 samples. 18 samples within our lung cancer panel were

re-typed by sequencing analysis. Besides the lung cancer samples, genomic DNA samples from

cancer cell lines were also typed with the same method. A number of inconsistencies remain

unresolved at presence. and are summarized in Table 21.

59

Table 21. Genotype typing inconsistencies

Genotype/Allele Experiment 1 Experiment 2

MTHFR 677

Sample #726 C/T (RFLP) C/C (RFLP)

Sample #745 C/T (RFLP) T/T (sequencing)

Sample #639 C/T (RFLP) C/C (sequencing)

MTRR 66

Sample #693 A/A (RFLP) G/G (RFLP)

RFC 80

SKCO1 cell line A/A (RFLP) A/G (sequencing)

For the MTHFR C677T genotype, with the RFLP analysis, we observed contradictory

results for 3 samples and typing with sequencing analysis generated inconsistent data for 2 out of

the 5 samples. There were 5 findings which we were unable to re-obtain by RFLP analysis for

the MTRR A66G polymorphism and 2 discordant results optained by sequencing for the RFC

G80A variant. These samples remain to be further analyzed.

The below experiment demonstrates and innate problem of RFLP, namely that

incomplete enzymatic digestion can cause erroneous results (Figure 4). The indicated Lung

cancer sampleswere subjected to Hinp1I digestion overnight, following PCR amplification. Lu-

183 DNA digestion generated a 154 bp band (as well as a smaller 124bp band not visible on the

figure) and therefore was classified as a GG homozygote. Lu 191 is a heterozygote, and so is Lu

726. However, despite the extended digestion time, Lu 718 and Lu 728 DNA digestion is most

likely incomplete.If incomplete digestion took place for the other samples as well, we could be

misinterpreting the results; for example all the above samples could be AA homozygotes.

Despite several repeats Lu 726 typing remains unresolved.

60

Figure 4. RFPL analysis of RFC G80A polymorphism.

Consequently, these results require us to reproduce our data by a more reliable and

ideally, a high-throughput method. One dependable and relatively new technique for SNP

analysis is the probe based real time quantitative polymerase chain reaction (Q-PCR) [72].

Q-PCR allows quantification of polymorphic DNA regions and genotyping of single

nucleotide polymorphisms in one run. The RFLP method, which involves 3 steps, requires the

amplification of the region around a given SNP and digestion of the PCR product by a restriction

enzyme with ensuing gel electrophoresis. However, with a single Q-PCR run, both quantification

and genotyping can be performed simultaneously. Online monitoring of the amplification

process as well as genotyping by melting curve analysis is possible by the use of hybridization

probes. The hybridization probes are sequence-specific oligonucleotides labeled by fluorescence

dyes. For the genotyping of the SNP, two hybridization probes are required. One covers the

polymorphic region (sensor hybridization probe) and carries a florescent dye that emits green

light. The second probe (the anchor probe) binds to a site close in proximity to the sensor

hybridization probe and carries a red light emitting dye. During the amplification process, the

hybridization probe anneals to the amplified DNA and emits green light, which then excites the

red dye in the anchor probe. The energy transferred from the sensor probe to the anchor probe is

called florescence resonance energy transfer (FRET). The FRET signal, which is detected by the

Q-PCR machine, is a direct measure of the DNA copy.

After amplification, the SNP’s can be detected by a melting-curve analysis. At this stage,

the temperature is raised from 40oC to 75oC. Then, the temperature at which the hybridization

61

probes are melted off the DNA strand (Tm), which is an indicator of the presence or the absence

of the mutation, is calculated. If the hybridization probe fits perfectly to the template DNA, then

it will have a higher Tm. If there is a single nucleotide mismatch, then the probe will melt at a

lower temperature. The melting curve analysis can yield three results. One is a curve with a

single early peak, second is a curve with a single late peak, and the last one is a curve with two

peaks. The results obtained by this method are quite reliable that the melting points for a specific

SNP is always at the same point. However, if there is another SNP within the region of the

hybridization probe binding site, then the melting points for each genotype will deviate. A

deviation more than 1.5oC in the melting curve step, is an indication of an additional mutation

within the binding region of the hybridization probe [73]. For instance, a recent study has shown

that an additional mutation causes deviations in the detection of the MTHFR A1298C SNP [72].

An exemplary graph shows the possible alleles and an unexpected nearby mutation curves

(Figure 5). Notice the heterozygote allele has a lower florescence value than the homozygote

alleles.

62

Figure 5. An exemplary RT-PCR melting-curve analysis showing the possible genotypes with an additional

unexpected mutation.

Thus quantitative Q-PCR offers several advantages over RFLP, However, given the fact

that there are a significantly large number of SNPs identified for each enzyme; sequencing

analysis would offer further reliability.

Despite the suggestion that enzyme digestion might be incomplete in a number of

experiments of ours, another possible explanation that can explain variations of band intensities

we consistently observed for some samples is that there might be a copy number variation

(CNV) in the genes we studied. What directed us to this conclusion was that upon the digestion

procedure of the MTRR 66 polymorphism, there was a greater intensity of the G allele than the

A allele in some of the heterozygote samples. This is demonstrated in Figure 6. The figure

demonstrates the RFLP results obtained for the MTRR A66G allele. The three bands obtained

for Lu 706 is equal in intensity and correspond to bands expected for a heterozygote (66, 44 and

22bp). The 44bp band is significantly more intense than the 66bp band. However, for Lu 706,

713, 716 and 658, the 66bp band is significantly more intense than the 44bp and 22bp bands. It is

unlikely that incomplete digestion accounts for these, because this experiment was repeated 4

times with equal results. If alleles are amplified in the genome resulting in CNV, then the

subsequent RFLP analysis would be expected to generate a result as shown in Figure 6.

63

Figure 6. A possible copy number variation in the MTRR A66G polymorphism. The “G” band is significantly

more intense than the “A” band in the 2nd, 3rd, 4th and 5th lanes whereas the intensity is equal in the 1st lane.

It is important to know whether there is a copy number variation (CNV) of these genes

and this should be investigated by the Q-PCR method explained above or CNV microarray

studies. The amplification step could reveal, the increase in the copy number of the one carbon

pathway genes. If the presence of a copy number variation in the MTRR gene and in other genes

is found, this possibly might have occurred towards the compensation of a reduced methyl group

flow in the one carbon pathway. For instance, if the patient has a MTHFR 677 hypoactive allele

(T), the cell could be amplifying the MTRR normoactive (C) allele in order to compensate for

methionine synthesis through an increased activation of MTR, in the presence of low production

of N5-methyl-THF caused by the MTHFR 677 T allele. Despite the fact that methionine

synthesis is N5-methyl-THF – dependent, the amplification might still have a functional

consequence. This could be novel a mechanism by which the cell maintains a specific rate of

SAM production in the presence of certain mutations as there are no publications documenting

CNV for these genes to date.

Our analysis of the correlations between the 1 –carbon pathway enzyme genotypes and

CT expression profiles of the lung cancer genomic DNA samples demonstrated important

contradictions. First of all, we found out that the CT (-) tumor samples had a higher proportion of

64

the MTHFR 677 T allele when compared to the CT (+) samples (see Table 21, Chapter 3). This

is difficult to reconcile with our hypothesis that the CT (+) samples have a higher proportion of

this allele if their expression is due to inefficient SAM production. Moreover, despite not being

statistically significant, 37.9% of the CT (-) were heterozygous for MTHFR 677, while only

28.6% of CT (+) samples were heterozygous.

In line with these results we also observed that for the MTRR A66G polymorphism.

MTRR 66 GG allele, thought to code for the hypoactive enzyme, was present among CT (-) lung

cancer samples at a rate that was five-fold more of what we observed for the CT (+) samples. As

explained above, because we hypothesized that the CT (+) would have the greater frequency of

hypoactive alleles, this data obtained appeared to be just the opposite of our initial hypothesis.

Our data demonstrates a strong negative correlation between MTHFR 677 CC and

MTRR 66 GG alleles, normoactive variants of these two enzymes (MTHFR 677 CC and MTRR

66 AG/AA); as well as between hypoactive variants (MTHFR 677 CT/TT and MTRR 66 GG).

This would further suggest that among lung cancer patients, there are two haplotypes, one prone

to efficient and the other prone to inefficient SAM production. Even though this is a plausible

possibility, it makes it even more difficult to explain why the inefficient SAM producers should

preferentially have CT (-) tumors. Therefore, I offer the following hypothesis that might possibly

help understand this dilemma.

It is known that SAM is used for the methylation of several substrates including RNA,

DNA and proteins. Some of the most important substrates of SAM are, therefore the histone

proteins. Histones are modified at specific residues in their N-terminal chains. These

modifications include phosphorylation, sumoylation, acetylation, ubiquitination as well as

methylation. The effect of each modification is being actively studied [74]. Methylation is one of

the most critical modifications regulating of histone function. Mono, di- or tri- methylation of

especially the 4th, 9th 27th and 79th lysine residues of H3 closely associate with repressed or

activated chromatin [75]. It has been shown that the modifications associated with active

chromatin include H3K4me2, HeK4me3, H3K36me3, while those associated with repressed

chromatin include H3K4me1, H3K36me2, H3K9me1 and H3K79me1 {Li, 2007}.

65

Methylthioadenosine (MTA), a nucleoside produced in the polyamine biosynthetic

pathway by the cleavage of SAM, is a known inhibitor of methyltransferases {Mato, 2007}.

MTA has been shown to inhibit H3K4 methylation [76], [77] by inhibiting Set1

methyltransferase [78]. In line with this observation, it has been shown that the addition of SAM

to the RAW cell line inhibited LPS-induced TNF-α gene expression in vitro {Ara, 2008}. In the

same study, it has also been observed by ChIP experiments that upon SAM treatment, H3K4 tri-

methylation is decreased compared to the LPS treated cells alone. H3K4 tri-methylation, as

explained before, is a hallmark of transcriptional activation. It has also been found that the

treatment of cells with SAM or MTA reduced the levels of H3K4me1 and H3K4me2, which

suggested that these agents might be playing a role in inhibiting the enzyme responsible for the

trimethylation of H3K4. Same observations were also obtained by a different study. Huang et al

have observed that MTA decreased the levels of both H3K4me2 and H3K4me3 [78]. MTA is

known to inhibit SAH hydrolase, the enzyme which converts SAH to homocysteine [79], and

SAH itself is a strong competitive inhibitor of almost all SAM-dependent methyltransferases

[82]. The SAH levels must be kept in check because many methyltransferases have a higher

affinity for SAH than SAM, and this makes SAH, as just mentioned, a potent inhibitor of many

methylation reactions [82].

Apart from affecting H3K4 methylation, MTA might also affects H3K9 methylation

since it is known that the G9a, the H3K9 methyltransferase is a SAM-dependent

methyltransferase. MTA, which reduces the level of H3K4 methylation, might also influence

H3K9 methylation. In addition to this, Mutskov et al conducted a study in which they showed

that H3K9 methylation can occur prior to DNA methylation [80].

Considering all of these observations, I have created a model in which CT expression is

primarily regulated by histone modifications, which are affected by SAM concentrations.

In this model, if the CT (-) lung cancer samples have a higher frequency of hypoactive

alleles of MTHFR 677 and MTRR 66, resulting in less efficient SAM production when

compared to CT (+) lung cancer samples. The inefficient production of SAM will yield

decreased amounts of its metabolite, MTA. The decreased amount of MTA will allow H3K4 and

66

H3K9 methylations which in turn result in DNA methylation and the CT genes will be silenced

(Figure 7).

On the other side, the wild-type alleles of MTHFR 677 and MTRR 66 genes will result in

an adequate production of SAM and consequently MTA. Because MTA is itself an inhibitor of

methyltransferases and SAH hydrolase, SAH levels will increase. Since SAH is a strong

inhibitor of almost all SAM dependent methyltransferases, along with MTA, they will have a

suppressive effect on histone methylation. This would result in a reduction of H3K4

trimethylation and H3K9 methylation. This will lead to disturbances in DNA methylation and

CT expression since H3K9 will not be methylated (Figure 7).

67

Mutant

MTHFR 677

RFC 80

SAM/SAH MTA

SAH

Set1

MT’s

H3K4

H3K9 methylationCT(-)

Wild-type

MTHFR 677

RFC 80

SAM/SAH MTA

SAH

Set1

MT’s

H3K4

H3K9 methylationCT(+)

Figure 7. Proposed model adopted from the results of this study for the regulation of CT genes

As a result, a new insight about CT gene expression could emerge. In order to test this

model, we first have to start by checking whether different histone methyltransferases have the

same sensitivity for SAM. This aspect of histone methyltransferases is important because

different sensitivities would mean that they will respond to different SAM concentrations by

distinct ways, which would eventually affect the activation/suppression of the target genes. There

is a procedure of determining the enzymatic activity by a method called peptide methylation

assay. In this assay described by Rathert et al [81], just like the procedure in ELISA experiments,

the wells on the plate are coated with a target peptide. After the addition of the desired

methyltransferases and SAM mixture into the wells, a continuous read-out of the reaction

process is performed. As the methyltransferase methylates the target peptide, light of a certain

68

wavelength is emitted and the amount of light is read and quantified. The time versus emission

graph will give us the kinetic activity of the enzyme. Then, ChIP experiments could be

performed in a given haplotype to see whether the polymorphisms in the one carbon enzyme

genes would have an effect on the histone modifications in CT genes.

In order to regard the effect of these polymorphisms thoroughly, we have to focus on

folic acid and DNA damage. As mentioned earlier, the deficiency of folic acid reduces the

substrate flow for the MTHFR enzyme and the SAM production is impaired due to inefficient

methionine production. Folic acid deficiency has been correlated with DNA hypomethylation.

On the other hand, thymidylate synthase prevents uracil misincorporation by converting uracil

into thymidine.

Another possible mechanism for the explanation of a model which could explain our

results belies on the assumption that different methyltransferases have different sensitivities to

SAM. In this model, because CT (-) samples have more frequent mutant genotypes, the SAM

production will be inefficient. If some methyltransferases are more sensitive to SAM

concentrations than others, the histone proteins might be preferentially modified. For instance, if

Set1, a H3K4 methyltransferase requires higher SAM concentrations for its optimum

performance than Ga9, a H3K9 methyltransferase, then preferential downregulation of H3K4

methylation, in low SAM concentrations, might result in a relatively high H3K9 methylation and

thus, repression of gene transcription. On the other hand, in CT (+) samples generating adequate

SAM concentrations, comparable H3K4- and H3K9-methyltransferase activities might result in

preferential H3K4 tri-methylation resulting in gene activation. Figure 8 summarizes this model.

69

Mutant

MTHFR 677

RFC 80

SAM/SAH

Wild-type

MTHFR 677

RFC 80

SAM/SAH

MT’s

MT’s

H3K4

H3K9

H3K4

H3K9

Figure 8. Proposed model for the regulation of CT genes adopted from the results of this study

For a better understanding of CT gene regulation and the effect of one carbon pathway

enzyme gene polymorphisms, we will pursue our studies by selecting two cell lines with a

certain a genotype which we know is associated with CT positivity. Then, by culturing these cell

lines in a folic acid deficient medium we’ll induce DNA damage. The DNA damage will be

evaluated by using the comet assay method, which is used for the detection of single and double

strand breaks on the DNA. The folic acid deficiency would result in decreased amounts of 5-

methylene-THF, which is a substrate for both MTHFR and TS. So, we would be able to see the

difference between a MTHFR 677 CC and MTHFR 677 TT genotype for the utilization of the 5-

70

methylene-THF in the methylation direction. The impaired MTHFR 677 TT genotypes are

expected to have a lesser degree of DNA damage because the thymidylate synthase enzyme will

have a larger amount of 5-methylene-THF for the conversion of dUMP to dTMP, preventing

uracil misincorporation. Moreover, by the addition of SAM into the cell culture, the expression

of CT genes will be observed.

In our subsequent studies, we might also focus on the methyltransferases and the

metabolism of SAM in order to further investigate the relationship between the production of

SAM, MTA and histone modifications. For this reason, we have pursued a preliminary micro-

array meta analysis from data compiled from 17 independent datasets corresponding to

microarray data obtained from different tumors.. Table 22 lists the genes which are significantly

up or downregulated, the probes and the number of experiments along with the average fold

change.

Table 22. The microarray meta-analysis showing the average fold changes of gene expression for the listed genes in cancer.

Gene Probe Direction Number of experiments

Average Fold change

TS 1 up 2 2.04 TS 2 up 11 3.44

MTR down 5 1.53

MTRR 1 down 4 1.77

MTRR 2 up 1 1.84

DHFR 1 up 2 1.67

DHFR 2 up 7 1.92

DHFR 3 up 6 1.96

DHFR 4 up 5 1.78

Methionine adenosyltransferase up 5 1.96

Adomet decarboxylase 1 up 7 2.68

Adomet decarboxylase 2 up 5 1.92

Spermine synthase up 3 1.76

Spermidine synthase up 8 1.98

Ornithine decarboxylase up 10 2.67

5’methyladenosine phosphorylase 1 up 1 1.65

5’methyladenosine phosphorylase 2 up 1 1.52

71

It can be inferred from the Table 16 that different enzymes which we are not focused on

might play important roles in SAM utilization and folate metabolism. The cancer cells may be

upregulating these genes in order to compensate the deficiency of their substrates. Along with

the thymidylate synthase expression data, obtained by Q-PCR, we can regard that different

cancer cell lines have diverse TS expression values. For instance, the LS174T cell line has

almost 6 times TS expression than the HCT15 cell line. These differences direct us to the one

carbon pathway enzyme gene polymorphisms and the expression levels of methyltransferases of

these two cell lines. These aspects of the pathway can be included in future studies.

Figure 9. Q-PCR results showing the expression levels of thymidylate synthase in different cancer cell

lines.

Differential TS expression in different cell lines might indicate a mechanism to

compensate for a hypoactive genotype of MTHFR 677 or severe DNA damage caused by uracil

72

misincorporation. Thus, as explained earlier [31], different cell lines might have distinct

mechanisms to maintain an efficient SAM production and DNA synthesis rates.

As a result, this study has shown that there might be a mechanistic bridge between the

production of SAM and DNA methylation, which affects cancer-testis gene expression. The

studies of folic acid intake, TS expression and activity should be pursued in order to regard their

effects on SAM production and especially histone methyltransferases. This study has helped us

gain new insight into the potential interplay between two mechanisms, the one carbon pathway

and epigenetic modification, and has made it possible to envision the correct experiments that

would be needed to elucidate this new aspect of CT gene regulation.

73

CHAPTER 5. REFERENCES

1. Kim, Y.I., Folate and DNA methylation: a mechanistic link between folate deficiency and

colorectal cancer? Cancer Epidemiol Biomarkers Prev, 2004. 13(4): p. 511-9.

2. Jaenisch, R. and A. Bird, Epigenetic regulation of gene expression: how the genome

integrates intrinsic and environmental signals. Nat Genet, 2003. 33 Suppl: p. 245-54.

3. Laird, P.W. and R. Jaenisch, The role of DNA methylation in cancer genetic and

epigenetics. Annu Rev Genet, 1996. 30: p. 441-64.

4. Turek-Plewa, J. and P.P. Jagodzinski, The role of mammalian DNA methyltransferases in

the regulation of gene expression. Cell Mol Biol Lett, 2005. 10(4): p. 631-47.

5. Regha, K., et al., Active and repressive chromatin are interspersed without spreading in

an imprinted gene cluster in the mammalian genome. Mol Cell, 2007. 27(3): p. 353-66.

6. Ginder, G.D., M.N. Gnanapragasam, and O.Y. Mian, The role of the epigenetic signal,

DNA methylation, in gene regulation during erythroid development. Curr Top Dev Biol,

2008. 82: p. 85-116.

7. Metivier, R., et al., Cyclical DNA methylation of a transcriptionally active promoter.

Nature, 2008. 452(7183): p. 45-50.

8. Margot, J.B., A.E. Ehrenhofer-Murray, and H. Leonhardt, Interactions within the

mammalian DNA methyltransferase family. BMC Mol Biol, 2003. 4: p. 7.

9. Brown, R. and G. Strathdee, Epigenomics and epigenetic therapy of cancer. Trends Mol

Med, 2002. 8(4 Suppl): p. S43-8.

10. Agrawal, A., R.F. Murphy, and D.K. Agrawal, DNA methylation in breast and colorectal

cancers. Mod Pathol, 2007. 20(7): p. 711-21.

11. Narayan, A., et al., Hypomethylation of pericentromeric DNA in breast

adenocarcinomas. Int J Cancer, 1998. 77(6): p. 833-8.

12. Hanahan, D. and R.A. Weinberg, The hallmarks of cancer. Cell, 2000. 100(1): p. 57-70.

13. Esteller, M., Cancer epigenetics: DNA methylation and chromatin alterations in human

cancer. Adv Exp Med Biol, 2003. 532: p. 39-49.

74

14. Guerrero-Preston, R., et al., Global DNA hypomethylation in liver cancer cases and

controls: a phase I preclinical biomarker development study. Epigenetics, 2007. 2(4): p.

223-6.

15. Chen, Y.T., et al., Identification of multiple cancer/testis antigens by allogeneic antibody

screening of a melanoma cell line library. Proc Natl Acad Sci U S A, 1998. 95(12): p.

6919-23.

16. Lengauer, C., K.W. Kinzler, and B. Vogelstein, DNA methylation and genetic instability

in colorectal cancer cells. Proc Natl Acad Sci U S A, 1997. 94(6): p. 2545-50.

17. Seng, T.J., et al., DLEC1 and MLH1 promoter methylation are associated with poor

prognosis in non-small cell lung carcinoma. Br J Cancer, 2008. 99(2): p. 375-82.

18. Kang, G.H., et al., Profile of aberrant CpG island methylation along multistep gastric

carcinogenesis. Lab Invest, 2003. 83(4): p. 519-26.

19. Kanai, Y. and S. Hirohashi, Alterations of DNA methylation associated with

abnormalities of DNA methyltransferases in human cancers during transition from a

precancerous to a malignant state. Carcinogenesis, 2007. 28(12): p. 2434-42.

20. Mohn, F., et al., Lineage-specific polycomb targets and de novo DNA methylation define

restriction and potential of neuronal progenitors. Mol Cell, 2008. 30(6): p. 755-66.

21. Tuck-Muller, C.M., et al., DNA hypomethylation and unusual chromosome instability in

cell lines from ICF syndrome patients. Cytogenet Cell Genet, 2000. 89(1-2): p. 121-8.

22. Scanlan, M.J., et al., Cancer/testis antigens: an expanding family of targets for cancer

immunotherapy. Immunol Rev, 2002. 188: p. 22-32.

23. Ross, M.T., et al., The DNA sequence of the human X chromosome. Nature, 2005.

434(7031): p. 325-37.

24. De Smet, C., et al., DNA methylation is the primary silencing mechanism for a set of

germ line- and tumor-specific genes with a CpG-rich promoter. Mol Cell Biol, 1999.

19(11): p. 7327-35.

25. Wischnewski, F., K. Pantel, and H. Schwarzenbach, Promoter demethylation and histone

acetylation mediate gene expression of MAGE-A1, -A2, -A3, and -A12 in human cancer

cells. Mol Cancer Res, 2006. 4(5): p. 339-49.

75

26. Gure, A.O., et al., Cancer-Testis Genes Are Coordinately Expressed and Are Markers of

Poor Outcome in Non-Small Cell Lung Cancer. Clin Cancer Res, 2005. 11(22): p. 8055-

8062.

27. Finkelstein, J.D., Metabolic regulatory properties of S-adenosylmethionine and S-

adenosylhomocysteine. Clin Chem Lab Med, 2007. 45(12): p. 1694-9.

28. Pogribny, I.P., et al., Irreversible global DNA hypomethylation as a key step in

hepatocarcinogenesis induced by dietary methyl deficiency. Mutat Res, 2006. 593(1-2):

p. 80-7.

29. Stempak, J.M., et al., Cell and stage of transformation-specific effects of folate deficiency

on methionine cycle intermediates and DNA methylation in an in vitro model.

Carcinogenesis, 2005. 26(5): p. 981-90.

30. Lucock, M., Folic acid: nutritional biochemistry, molecular biology, and role in disease

processes. Mol Genet Metab, 2000. 71(1-2): p. 121-38.

31. Hayashi, I., et al., Folate deficiency induces cell-specific changes in the steady-state

transcript levels of genes involved in folate metabolism and 1-carbon transfer reactions

in human colonic epithelial cells. J Nutr, 2007. 137(3): p. 607-13.

32. Selhub, J., Homocysteine metabolism. Annu Rev Nutr, 1999. 19: p. 217-46.

33. Huang, S., Histone methyltransferases, diet nutrients and tumour suppressors. Nat Rev

Cancer, 2002. 2(6): p. 469-76.

34. Xu, X., et al., Polymorphisms of one-carbon-metabolizing genes and risk of breast

cancer in a population-based study. Carcinogenesis, 2007. 28(7): p. 1504-9.

35. Blount, B.C., et al., Folate deficiency causes uracil misincorporation into human DNA

and chromosome breakage: implications for cancer and neuronal damage. Proc Natl

Acad Sci U S A, 1997. 94(7): p. 3290-5.

36. Wickramasinghe, S.N. and S. Fida, Bone marrow cells from vitamin B12- and folate-

deficient patients misincorporate uracil into DNA. Blood, 1994. 83(6): p. 1656-61.

37. Crott, J.W., et al., Moderate folate depletion modulates the expression of selected genes

involved in cell cycle, intracellular signaling and folate uptake in human colonic

epithelial cell lines. J Nutr Biochem, 2008. 19(5): p. 328-35.

76

38. Branda, R.F., et al., Effects of folate deficiency on the metastatic potential of murine

melanoma cells. Cancer Res, 1988. 48(16): p. 4529-34.

39. Melnyk, S., et al., Uracil misincorporation, DNA strand breaks, and gene amplification

are associated with tumorigenic cell transformation in folate deficient/repleted Chinese

hamster ovary cells. Cancer Lett, 1999. 146(1): p. 35-44.

40. Rodriguez, J., et al., Chromosomal instability correlates with genome-wide DNA

demethylation in human primary colorectal cancers. Cancer Res, 2006. 66(17): p. 8462-

9468.

41. Yuan, J.M., et al., Genetic polymorphisms in the methylenetetrahydrofolate reductase

and thymidylate synthase genes and risk of hepatocellular carcinoma. Hepatology, 2007.

46(3): p. 749-58.

42. Martin, Y.N., et al., Human methylenetetrahydrofolate reductase pharmacogenomics:

gene resequencing and functional genomics. Pharmacogenet Genomics, 2006. 16(4): p.

265-77.

43. Frosst, P., et al., A candidate genetic risk factor for vascular disease: a common mutation

in methylenetetrahydrofolate reductase. Nat Genet, 1995. 10(1): p. 111-3.

44. Heijmans, B.T., et al., A common variant of the methylenetetrahydrofolate reductase

gene (1p36) is associated with an increased risk of cancer. Cancer Res, 2003. 63(6): p.

1249-53.

45. Castro, R., et al., 5,10-methylenetetrahydrofolate reductase (MTHFR) 677C-->T and

1298A-->C mutations are associated with DNA hypomethylation. J Med Genet, 2004.

41(6): p. 454-8.

46. Friso, S., et al., The MTHFR 1298A>C polymorphism and genomic DNA methylation in

human lymphocytes. Cancer Epidemiol Biomarkers Prev, 2005. 14(4): p. 938-43.

47. Botto, N., et al., Genetic polymorphisms in folate and homocysteine metabolism as risk

factors for DNA damage. Eur J Hum Genet, 2003. 11(9): p. 671-8.

48. Sharp, L. and J. Little, Polymorphisms in genes involved in folate metabolism and

colorectal neoplasia: a HuGE review. Am J Epidemiol, 2004. 159(5): p. 423-43.

77

49. Stern, L.L., et al., Genomic DNA hypomethylation, a characteristic of most cancers, is

present in peripheral leukocytes of individuals who are homozygous for the C677T

polymorphism in the methylenetetrahydrofolate reductase gene. Cancer Epidemiol

Biomarkers Prev, 2000. 9(8): p. 849-53.

50. Relton, C.L., et al., Gene-gene interaction in folate-related genes and risk of neural tube

defects in a UK population. J Med Genet, 2004. 41(4): p. 256-60.

51. Olteanu, H., T. Munson, and R. Banerjee, Differences in the efficiency of reductive

activation of methionine synthase and exogenous electron acceptors between the common

polymorphic variants of human methionine synthase reductase. Biochemistry, 2002.

41(45): p. 13378-85.

52. Devlin, A.M., et al., Effect of Mthfr genotype on diet-induced hyperhomocysteinemia and

vascular function in mice. Blood, 2004. 103(7): p. 2624-9.

53. Christensen, B., et al., Correlation of a common mutation in the

methylenetetrahydrofolate reductase gene with plasma homocysteine in patients with

premature coronary artery disease. Arterioscler Thromb Vasc Biol, 1997. 17(3): p. 569-

73.

54. Leclerc, D., et al., Human methionine synthase: cDNA cloning and identification of

mutations in patients of the cblG complementation group of folate/cobalamin disorders.

Hum Mol Genet, 1996. 5(12): p. 1867-74.

55. Chango, A., et al., A polymorphism (80G->A) in the reduced folate carrier gene and its

associations with folate status and homocysteinemia. Mol Genet Metab, 2000. 70(4): p.

310-5.

56. Whetstine, J.R., et al., Single nucleotide polymorphisms in the human reduced folate

carrier: characterization of a high-frequency G/A variant at position 80 and transport

properties of the His(27) and Arg(27) carriers. Clin Cancer Res, 2001. 7(11): p. 3416-22.

57. Horie, N., et al., Functional analysis and DNA polymorphism of the tandemly repeated

sequences in the 5'-terminal regulatory region of the human gene for thymidylate

synthase. Cell Struct Funct, 1995. 20(3): p. 191-7.

78

58. Trinh, B.N., et al., Thymidylate synthase: a novel genetic determinant of plasma

homocysteine and folate levels. Hum Genet, 2002. 111(3): p. 299-302.

59. Suzuki, T., et al., Impact of one-carbon metabolism-related gene polymorphisms on risk

of lung cancer in Japan: a case control study. Carcinogenesis, 2007. 28(8): p. 1718-25.

60. Ulrich, C.M., et al., Polymorphisms in the reduced folate carrier, thymidylate synthase,

or methionine synthase and risk of colon cancer. Cancer Epidemiol Biomarkers Prev,

2005. 14(11 Pt 1): p. 2509-16.

61. Markus, H.S., et al., A common polymorphism in the methylenetetrahydrofolate reductase

gene, homocysteine, and ischemic cerebrovascular disease. Stroke, 1997. 28(9): p. 1739-

43.

62. Paz, M.F., et al., Germ-line variants in methyl-group metabolism genes and susceptibility

to DNA methylation in normal tissues and human primary tumors. Cancer Res, 2002.

62(15): p. 4519-24.

63. Friso, S., et al., A common mutation in the 5,10-methylenetetrahydrofolate reductase

gene affects genomic DNA methylation through an interaction with folate status. Proc

Natl Acad Sci U S A, 2002. 99(8): p. 5606-11.

64. Vaughn, J.D., et al., Methionine synthase reductase 66A->G polymorphism is associated

with increased plasma homocysteine concentration when combined with the homozygous

methylenetetrahydrofolate reductase 677C->T variant. J Nutr, 2004. 134(11): p. 2985-

90.

65. Moore, L.E., et al., Folate metabolism genes, vegetable intake and renal cancer risk in

central Europe. Int J Cancer, 2008. 122(8): p. 1710-5.

66. Robertson, K.D. and A.P. Wolffe, DNA methylation in health and disease. Nat Rev

Genet, 2000. 1(1): p. 11-9.

67. Jones, P.A. and J.D. Buckley, The role of DNA methylation in cancer. Adv Cancer Res,

1990. 54: p. 1-23.

68. del Mazo, J., et al., DNA methylation changes during mouse spermatogenesis.

Chromosome Res, 1994. 2(2): p. 147-52.

79

69. Shi, Q., et al., Sex differences in risk of lung cancer associated with methylene-

tetrahydrofolate reductase polymorphisms. Cancer Epidemiol Biomarkers Prev, 2005.

14(6): p. 1477-84.

70. Pfaffl, M.W., A new mathematical model for relative quantification in real-time RT-PCR.

Nucleic Acids Res, 2001. 29(9): p. e45.

71. Tanaka, S., et al., O6-methylguanine-DNA methyltranspherase gene expression in

gliomas by means of real-time quantitative RT-PCR and clinical response to

nitrosoureas. Int J Cancer, 2003. 103(1): p. 67-72.

72. Allen, R.A., et al., A common 1317TC polymorphism in MTHFR can lead to erroneous

1298AC genotyping by PCR-RE and TaqMan probe assays. Genet Test, 2007. 11(2): p.

167-73.

73. Reuter, M., et al., Detection of new single nucleotide polymorphisms by means of real

time PCR. J Genet, 2005. 84(3): p. 341-5.

74. Kouzarides, T., Chromatin modifications and their function. Cell, 2007. 128(4): p. 693-

705.

75. Barski, A., et al., High-resolution profiling of histone methylations in the human genome.

Cell, 2007. 129(4): p. 823-37.

76. Chau, C.M. and P.M. Lieberman, Dynamic chromatin boundaries delineate a latency

control region of Epstein-Barr virus. J Virol, 2004. 78(22): p. 12308-19.

77. Song, M.R. and A. Ghosh, FGF2-induced chromatin remodeling regulates CNTF-

mediated gene expression and astrocyte differentiation. Nat Neurosci, 2004. 7(3): p. 229-

35.

78. Huang, J., et al., Trimethylation of histone H3 lysine 4 by Set1 in the lytic infection of

human herpes simplex virus 1. J Virol, 2006. 80(12): p. 5740-6.

79. Ferro, A.J., A.A. Vandenbark, and M.R. MacDonald, Inactivation of S-

adenosylhomocysteine hydrolase by 5'-deoxy-5'-methylthioadenosine. Biochem Biophys

Res Commun, 1981. 100(2): p. 523-31.

80. Mutskov, V. and G. Felsenfeld, Silencing of transgene transcription precedes

methylation of promoter DNA and histone H3 lysine 9. Embo J, 2004. 23(1): p. 138-49.

80

81. Rathert, P., X. Cheng, and A. Jeltsch, Continuous enzymatic assay for histone lysine

methyltransferases. Biotechniques, 2007. 43(5): p. 602, 604, 606 passim.

82. Clarke S, Banfield K. S-adenosylmethionine-dependent methyltransferases: potential

targets in homocysteine-linked pathology. In: Carmel R, Jacobsen D, eds. Homocysteine

in Health and Disease. Cambridge, England: Cambridge University Press; 2001:63-78.

83. Hosmer W, Lameshow S. Applied Logistic Regression. Wiley Series; 2000

84. Davis, C. D. and E. O. Uthus (2004). "DNA methylation, cancer susceptibility, and

nutrient interactions." Exp Biol Med (Maywood) 229(10): 988-95.

81

SUPPLEMENTARY FIGURES

Table 23 (Supplementary Table 1). The genotype data of the CT (+) lung cancer patients

Lu MTHFR C677T MTHFR A1298C MTR A2756G MTRR A66G RFC G80A

31 CT AC AA AG AA

68 CT AA AG GG AG

87 CC AC AA GG GG

89 CT AC AG AG AG

111 CC AC AA AA GG

131 CC AA AA AG GG

168 CC AA AG AG AG

185 CT AA AG AG AG

186 CC AC AA AG GG

219 CT AA AA GG GG

223 TT AA AA GG AG

649 CC CC AG GG AA

652 CC AA AG AG AA

658 CC AC AG AG AG

726 CC AC AA AG AG

736 CC AA AG AG AG

739 TT AA AG GG AG

745 CT AC AA GG GG

752 CC AA AA AG AA

753 CC AA AG AG AA

759 CT AA AG AA AG

82

Table 24 (Supplementary Table 2). The genotype data of the CT (-) lung cancer patients

Lu MTHFR C677T MTHFR A1298C MTR A2756G MTRR A66G RFC G80A

69 CT AC AA GG AA

77 TT AA AA AG GG

88 CT AA AG GG GG

90 CT AA AA AA AG

108 CC AC AG AG AG

110 CT CC AA GG AA

112 CT AC AA GG AG

180 CT AC AG AA AG

183 CT AC AA AG AA

191 CC AC AA AG AG

221 TT AA AG GG GG

225 CC AC AA AA AG

639 CT AC AA GG

656 TT AA AA GG

670 TT AA AG AG AG

692 CT AA AG AG

693 TT AC AG GG AG

694 TT AC AG AG

698 CC AC AG AG AG

706 CC AA AG AA

707 CT AC AG AG AA

713 CC AC AG AG AG

716 TT AC AG AG AG

718 CT AA AA AA AG

728 CC AC AA AA AG

748 CT AA AA GG AG

749 CC AA AA AG AG

751 TT AA AA GG AG

763 CC CC AA GG AA

83

Figure 10 (Supplementary Figure 1). RFLP results of the MTHFR C677T polymorphisms in lung cancer

patients.

84

Figure 11 (Supplementary Figure 2). RFLP results of the MTHFR A1298C polymorphisms in lung cancer

patients.

85

Figure 12 (Supplementary Figure 3). RFLP results of the MTR A2756G polymorphisms in lung cancer patients

86

Figure 13 (Supplementary Figure 4). RFLP results of the MTRR A66G polymorphisms in lung cancer

patients.

87

Supplementary Figure 5. RFLP results of the RFC G80A polymorphisms in lung cancer patients.